Warning: Permanently added '174.129.140.177' (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/10237840-fedora-rawhide-x86_64 --chroot fedora-rawhide-x86_64


Version: 1.6
PID: 12428
Logging PID: 12430
Task:
{'allow_user_ssh': False,
 'appstream': False,
 'background': True,
 'build_id': 10237840,
 'buildroot_pkgs': [],
 'chroot': 'fedora-rawhide-x86_64',
 'enable_net': False,
 'fedora_review': False,
 'git_hash': 'a6b26681bf7591064156a87f23a4f2f5adcaa314',
 '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': 'b8064-1',
 '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': '10237840-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-uie1_oaj/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-uie1_oaj/llama-cpp', '--depth', '500', '--no-single-branch', '--recursive']
cwd: .
rc: 0
stdout: 
stderr: Cloning into '/var/lib/copr-rpmbuild/workspace/workdir-uie1_oaj/llama-cpp'...

Running: git checkout a6b26681bf7591064156a87f23a4f2f5adcaa314 --

cmd: ['git', 'checkout', 'a6b26681bf7591064156a87f23a4f2f5adcaa314', '--']
cwd: /var/lib/copr-rpmbuild/workspace/workdir-uie1_oaj/llama-cpp
rc: 0
stdout: 
stderr: Note: switching to 'a6b26681bf7591064156a87f23a4f2f5adcaa314'.

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 a6b2668 automatic import of llama-cpp

Running: dist-git-client sources

cmd: ['dist-git-client', 'sources']
cwd: /var/lib/copr-rpmbuild/workspace/workdir-uie1_oaj/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-b8064.tar.gz
INFO: Reading stdout from command: curl --help all
INFO: Calling: curl -H Pragma: -H 'Accept-Encoding: identity' -o llama.cpp-b8064.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-b8064.tar.gz/md5/e7236a822fccbfcdd62b7b897eb20975/llama.cpp-b8064.tar.gz
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100 27.6M  100 27.6M    0     0   448M      0 --:--:-- --:--:-- --:--:--  453M
INFO: Reading stdout from command: md5sum llama.cpp-b8064.tar.gz

tail: /var/lib/copr-rpmbuild/main.log: file truncated
Running (timeout=18000): unbuffer mock --spec /var/lib/copr-rpmbuild/workspace/workdir-uie1_oaj/llama-cpp/llama-cpp.spec --sources /var/lib/copr-rpmbuild/workspace/workdir-uie1_oaj/llama-cpp --resultdir /var/lib/copr-rpmbuild/results --uniqueext 1773846985.598827 -r /var/lib/copr-rpmbuild/results/configs/child.cfg
INFO: mock.py version 6.7 starting (python version = 3.14.2, NVR = mock-6.7-1.fc43), args: /usr/libexec/mock/mock --spec /var/lib/copr-rpmbuild/workspace/workdir-uie1_oaj/llama-cpp/llama-cpp.spec --sources /var/lib/copr-rpmbuild/workspace/workdir-uie1_oaj/llama-cpp --resultdir /var/lib/copr-rpmbuild/results --uniqueext 1773846985.598827 -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-uie1_oaj/llama-cpp/llama-cpp.spec)  Config(fedora-rawhide-x86_64)
Start: clean chroot
Finish: clean chroot
Mock Version: 6.7
INFO: Mock Version: 6.7
Start(bootstrap): chroot init
INFO: mounting tmpfs at /var/lib/mock/fedora-rawhide-x86_64-bootstrap-1773846985.598827/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-cf9e075d-e053-4081-9f41-f29a1fa3a7df
INFO: Checking that 2ed347e21456ef1ee79a9f817a127689cd8403f688fbe2c3466c26b791606d24 image matches host's architecture
INFO: Copy content of container 2ed347e21456ef1ee79a9f817a127689cd8403f688fbe2c3466c26b791606d24 to /var/lib/mock/fedora-rawhide-x86_64-bootstrap-1773846985.598827/root
INFO: mounting 2ed347e21456ef1ee79a9f817a127689cd8403f688fbe2c3466c26b791606d24 with podman image mount
INFO: image 2ed347e21456ef1ee79a9f817a127689cd8403f688fbe2c3466c26b791606d24 as /var/lib/containers/storage/overlay/bb83824bc629f9362f291ee2ccc807dfd9b852419988a804915cdf28e2acaec5/merged
INFO: umounting image 2ed347e21456ef1ee79a9f817a127689cd8403f688fbe2c3466c26b791606d24 (/var/lib/containers/storage/overlay/bb83824bc629f9362f291ee2ccc807dfd9b852419988a804915cdf28e2acaec5/merged) with podman image umount
INFO: Removing image mock-bootstrap-cf9e075d-e053-4081-9f41-f29a1fa3a7df
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_64-1773846985.598827/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-6.0.1-5.fc45.x86_64
  rpm-sequoia-1.10.1-1.fc45.x86_64
  dnf5-5.4.0.0-5.fc45.x86_64
  dnf5-plugins-5.4.0.0-5.fc45.x86_64
Start: installing minimal buildroot with dnf5
Updating and loading repositories:
 Copr repository                        100% | 770.6 KiB/s | 245.8 KiB |  00m00s
 fedora                                 100% |  20.2 MiB/s |  21.8 MiB |  00m01s
Repositories loaded.
Package                            Arch   Version                      Repository      Size
Installing group/module packages:
 bash                              x86_64 0:5.3.9-3.fc44               fedora       8.5 MiB
 bzip2                             x86_64 0:1.0.8-23.fc44              fedora      95.0 KiB
 coreutils                         x86_64 0:9.10-3.fc45                fedora       5.6 MiB
 cpio                              x86_64 0:2.15-9.fc44                fedora       1.1 MiB
 diffutils                         x86_64 0:3.12-5.fc44                fedora       1.6 MiB
 fedora-release-common             noarch 0:45-0.2                     fedora      20.6 KiB
 findutils                         x86_64 1:4.10.0-7.fc44              fedora       1.9 MiB
 gawk                              x86_64 0:5.4.0-2.fc45               fedora       2.2 MiB
 glibc-minimal-langpack            x86_64 0:2.43.9000-4.fc45           fedora       0.0   B
 grep                              x86_64 0:3.12-3.fc44                fedora       1.0 MiB
 gzip                              x86_64 0:1.14-2.fc44                fedora     401.6 KiB
 info                              x86_64 0:7.3-1.fc45                 fedora     372.4 KiB
 patch                             x86_64 0:2.8-4.fc44                 fedora     226.6 KiB
 redhat-rpm-config                 noarch 0:344-1.fc45                 fedora     183.7 KiB
 rpm-build                         x86_64 0:6.0.1-5.fc45               fedora     294.6 KiB
 sed                               x86_64 0:4.9-8.fc45                 fedora     857.2 KiB
 shadow-utils                      x86_64 2:4.19.3-1.fc45              fedora       4.0 MiB
 tar                               x86_64 2:1.35-8.fc44                fedora       3.0 MiB
 unzip                             x86_64 0:6.0-69.fc44                fedora     445.8 KiB
 util-linux                        x86_64 0:2.41.3-12.fc44             fedora       3.5 MiB
 which                             x86_64 0:2.23-4.fc44                fedora      83.4 KiB
 xz                                x86_64 1:5.8.2-2.fc44               fedora       1.3 MiB
Installing dependencies:
 R-srpm-macros                     noarch 0:1.3.5-1.fc45               fedora       3.5 KiB
 add-determinism                   x86_64 0:0.7.3-1.fc45               fedora       2.2 MiB
 alternatives                      x86_64 0:1.33-5.fc44                fedora      62.1 KiB
 ansible-srpm-macros               noarch 0:1-20.1.fc44                fedora      35.7 KiB
 audit-libs                        x86_64 0:4.1.3-1.fc44               fedora     390.5 KiB
 binutils                          x86_64 0:2.46.50-3.fc45             copr_base   28.3 MiB
 build-reproducibility-srpm-macros noarch 0:0.7.3-1.fc45               fedora       1.2 KiB
 bzip2-libs                        x86_64 0:1.0.8-23.fc44              fedora      80.5 KiB
 ca-certificates                   noarch 0:2025.2.80_v9.0.304-6.fc45  fedora       2.7 MiB
 cmake-srpm-macros                 noarch 0:4.2.3-2.fc45               fedora     524.0   B
 coreutils-common                  x86_64 0:9.10-3.fc45                fedora      10.7 MiB
 crypto-policies                   noarch 0:20251128-3.git19878fe.fc44 fedora     132.6 KiB
 curl                              x86_64 0:8.19.0~rc3-1.fc45          fedora     483.4 KiB
 cyrus-sasl-lib                    x86_64 0:2.1.28-35.fc44             fedora       2.3 MiB
 debugedit                         x86_64 0:5.3-2.fc45                 fedora     220.8 KiB
 dwz                               x86_64 0:0.16-3.fc44                fedora     290.9 KiB
 ed                                x86_64 0:1.22.5-2.fc45              fedora     149.7 KiB
 efi-srpm-macros                   noarch 0:6-6.fc44                   fedora      40.2 KiB
 elfutils                          x86_64 0:0.194-5.fc45               fedora       2.9 MiB
 elfutils-debuginfod-client        x86_64 0:0.194-5.fc45               fedora      83.7 KiB
 elfutils-default-yama-scope       noarch 0:0.194-5.fc45               fedora       1.8 KiB
 elfutils-libelf                   x86_64 0:0.194-5.fc45               fedora       1.1 MiB
 elfutils-libs                     x86_64 0:0.194-5.fc45               fedora     715.3 KiB
 erlang-srpm-macros                noarch 0:0.3.11-1.fc45              fedora       1.9 KiB
 fedora-gpg-keys                   noarch 0:45-0.1                     fedora     133.4 KiB
 fedora-release                    noarch 0:45-0.2                     fedora       0.0   B
 fedora-release-identity-basic     noarch 0:45-0.2                     fedora     664.0   B
 fedora-repos                      noarch 0:45-0.1                     fedora       4.9 KiB
 fedora-repos-rawhide              noarch 0:45-0.1                     fedora       2.2 KiB
 file                              x86_64 0:5.47-1.fc45                fedora     101.2 KiB
 file-libs                         x86_64 0:5.47-1.fc45                fedora      12.2 MiB
 filesystem                        x86_64 0:3.18-56.fc45               fedora     112.0   B
 filesystem-srpm-macros            noarch 0:3.18-56.fc45               fedora      38.2 KiB
 fonts-srpm-macros                 noarch 1:5.0.0-3.fc45               fedora      55.8 KiB
 forge-srpm-macros                 noarch 0:0.4.0-4.fc44               fedora      38.9 KiB
 fpc-srpm-macros                   noarch 0:1.3-16.fc44                fedora     144.0   B
 gap-srpm-macros                   noarch 0:2-2.fc44                   fedora       2.1 KiB
 gdb-minimal                       x86_64 0:17.1-5.fc45                fedora      14.2 MiB
 gdbm-libs                         x86_64 1:1.23-11.fc44               fedora     129.6 KiB
 ghc-srpm-macros                   noarch 0:1.10-1.fc44                fedora     792.0   B
 glibc                             x86_64 0:2.43.9000-4.fc45           fedora       7.0 MiB
 glibc-common                      x86_64 0:2.43.9000-4.fc45           fedora       1.0 MiB
 glibc-gconv-extra                 x86_64 0:2.43.9000-4.fc45           fedora       7.7 MiB
 gmp                               x86_64 1:6.3.0-5.fc44               fedora     815.2 KiB
 gnat-srpm-macros                  noarch 0:7-2.fc44                   fedora       1.0 KiB
 gnulib-l10n                       noarch 0:20241231-2.fc44            fedora     655.0 KiB
 gnupg2                            x86_64 0:2.4.9-5.fc44               fedora       6.5 MiB
 gnupg2-dirmngr                    x86_64 0:2.4.9-5.fc44               fedora     633.9 KiB
 gnupg2-gpg-agent                  x86_64 0:2.4.9-5.fc44               fedora     682.5 KiB
 gnupg2-gpgconf                    x86_64 0:2.4.9-5.fc44               fedora     249.7 KiB
 gnupg2-keyboxd                    x86_64 0:2.4.9-5.fc44               fedora     201.2 KiB
 gnupg2-verify                     x86_64 0:2.4.9-5.fc44               fedora     360.3 KiB
 gnutls                            x86_64 0:3.8.12-1.fc45              fedora       3.7 MiB
 go-srpm-macros                    noarch 0:3.8.0-2.fc44               fedora      61.9 KiB
 gpgverify                         noarch 0:2.2-4.fc44                 fedora       8.7 KiB
 ima-evm-utils-libs                x86_64 0:1.6.2-10.fc45              fedora      60.6 KiB
 jansson                           x86_64 0:2.14-4.fc44                fedora      88.9 KiB
 java-srpm-macros                  noarch 0:1-8.fc44                   fedora     870.0   B
 json-c                            x86_64 0:0.18-8.fc44                fedora      82.6 KiB
 kernel-srpm-macros                noarch 0:1.0-28.fc44                fedora       1.9 KiB
 keyutils-libs                     x86_64 0:1.6.3-7.fc44               fedora      54.2 KiB
 krb5-libs                         x86_64 0:1.22.2-2.fc45              fedora       2.4 MiB
 libacl                            x86_64 0:2.3.2-6.fc44               fedora      35.8 KiB
 libarchive                        x86_64 0:3.8.6-1.fc45               fedora     995.2 KiB
 libassuan                         x86_64 0:2.5.7-5.fc44               fedora     163.8 KiB
 libattr                           x86_64 0:2.5.2-8.fc44               fedora      24.3 KiB
 libblkid                          x86_64 0:2.41.3-12.fc44             fedora     274.3 KiB
 libbrotli                         x86_64 0:1.2.0-3.fc44               fedora     865.0 KiB
 libcap                            x86_64 0:2.77-2.fc44                fedora     212.1 KiB
 libcap-ng                         x86_64 0:0.9.1-1.fc45               fedora      68.7 KiB
 libcbor                           x86_64 0:0.13.0-2.fc44              fedora      79.5 KiB
 libcom_err                        x86_64 0:1.47.4-1.fc45              fedora      63.0 KiB
 libcurl                           x86_64 0:8.19.0~rc3-1.fc45          fedora       1.0 MiB
 libeconf                          x86_64 0:0.7.9-3.fc44               fedora      64.8 KiB
 libevent                          x86_64 0:2.1.12-17.fc44             fedora     978.7 KiB
 libfdisk                          x86_64 0:2.41.3-12.fc44             fedora     388.3 KiB
 libffi                            x86_64 0:3.5.2-2.fc44               fedora      87.7 KiB
 libfido2                          x86_64 0:1.16.0-5.fc44              fedora     246.4 KiB
 libfsverity                       x86_64 0:1.7-1.fc45                 fedora      28.4 KiB
 libgcc                            x86_64 0:16.0.1-0.9.fc45            copr_base  270.7 KiB
 libgcrypt                         x86_64 0:1.12.1-1.fc45              fedora       1.7 MiB
 libgomp                           x86_64 0:16.0.1-0.9.fc45            copr_base  572.7 KiB
 libgpg-error                      x86_64 0:1.59-1.fc45                fedora     946.4 KiB
 libidn2                           x86_64 0:2.3.8-3.fc44               fedora     556.4 KiB
 libksba                           x86_64 0:1.6.8-1.fc45               fedora     421.9 KiB
 liblastlog2                       x86_64 0:2.41.3-12.fc44             fedora      37.6 KiB
 libmount                          x86_64 0:2.41.3-12.fc44             fedora     388.6 KiB
 libnghttp2                        x86_64 0:1.68.0-3.fc44              fedora     166.1 KiB
 libnghttp3                        x86_64 0:1.15.0-1.fc44              fedora     159.2 KiB
 libpkgconf                        x86_64 0:2.5.1-1.fc45               fedora      90.1 KiB
 libpsl                            x86_64 0:0.21.5-7.fc44              fedora      76.3 KiB
 libselinux                        x86_64 0:3.10-1.fc44                fedora     201.0 KiB
 libselinux-utils                  x86_64 0:3.10-1.fc44                fedora     305.7 KiB
 libsemanage                       x86_64 0:3.10-1.fc44                fedora     312.3 KiB
 libsepol                          x86_64 0:3.10-1.fc44                fedora     870.0 KiB
 libsmartcols                      x86_64 0:2.41.3-12.fc44             fedora     188.3 KiB
 libssh                            x86_64 0:0.12.0-1.fc45              fedora     719.1 KiB
 libssh-config                     noarch 0:0.12.0-1.fc45              fedora     277.0   B
 libstdc++                         x86_64 0:16.0.1-0.9.fc45            copr_base    2.9 MiB
 libtasn1                          x86_64 0:4.21.0-1.fc45              fedora     180.6 KiB
 libtool-ltdl                      x86_64 0:2.5.4-10.fc44              fedora      70.0 KiB
 libunistring                      x86_64 0:1.1-11.fc44                fedora       1.7 MiB
 libusb1                           x86_64 0:1.0.29-5.fc44              fedora     175.2 KiB
 libuuid                           x86_64 0:2.41.3-12.fc44             fedora      37.2 KiB
 libverto                          x86_64 0:0.3.2-12.fc44              fedora      25.3 KiB
 libxcrypt                         x86_64 0:4.5.2-3.fc44               fedora     293.2 KiB
 libxml2                           x86_64 0:2.12.10-6.fc44             fedora       1.8 MiB
 libzstd                           x86_64 0:1.5.7-5.fc44               fedora     956.1 KiB
 linkdupes                         x86_64 0:0.7.3-1.fc45               fedora     780.9 KiB
 lua-libs                          x86_64 0:5.5.0-1.fc45               fedora     297.9 KiB
 lua-srpm-macros                   noarch 0:1-17.fc44                  fedora       1.3 KiB
 lz4-libs                          x86_64 0:1.10.0-4.fc44              fedora     157.3 KiB
 mpfr                              x86_64 0:4.2.2-3.fc44               fedora     849.1 KiB
 ncurses-base                      noarch 0:6.6-1.fc44                 fedora     329.7 KiB
 ncurses-libs                      x86_64 0:6.6-1.fc44                 fedora     968.9 KiB
 nettle                            x86_64 0:3.10.1-3.fc44              fedora     794.3 KiB
 ngtcp2                            x86_64 0:1.19.0-2.fc44              fedora     330.2 KiB
 ngtcp2-crypto-ossl                x86_64 0:1.19.0-2.fc44              fedora      51.6 KiB
 npth                              x86_64 0:1.8-4.fc44                 fedora      49.5 KiB
 ocaml-srpm-macros                 noarch 0:11-3.fc44                  fedora       1.9 KiB
 openblas-srpm-macros              noarch 0:2-21.fc44                  fedora     112.0   B
 openldap                          x86_64 0:2.6.13-1.fc45              fedora     669.9 KiB
 openssl-libs                      x86_64 1:3.5.5-1.fc44               fedora       9.2 MiB
 p11-kit                           x86_64 0:0.26.2-1.fc45              fedora       2.6 MiB
 p11-kit-trust                     x86_64 0:0.26.2-1.fc45              fedora     478.3 KiB
 package-notes-srpm-macros         noarch 0:0.17-3.fc45                fedora       1.6 KiB
 pam-libs                          x86_64 0:1.7.2-1.fc44               fedora     130.5 KiB
 pcre2                             x86_64 0:10.47-1.fc44.1             fedora     718.6 KiB
 pcre2-syntax                      noarch 0:10.47-1.fc44.1             fedora     281.9 KiB
 perl-srpm-macros                  noarch 0:1-61.fc44                  fedora     861.0   B
 pkgconf                           x86_64 0:2.5.1-1.fc45               fedora      92.7 KiB
 pkgconf-m4                        noarch 0:2.5.1-1.fc45               fedora      14.3 KiB
 pkgconf-pkg-config                x86_64 0:2.5.1-1.fc45               fedora     989.0   B
 policycoreutils                   x86_64 0:3.10-2.fc45                fedora     884.4 KiB
 popt                              x86_64 0:1.19-10.fc44               fedora     132.6 KiB
 publicsuffix-list-dafsa           noarch 0:20260116-1.fc44            fedora      70.4 KiB
 pyproject-srpm-macros             noarch 0:1.18.7-1.fc45              fedora       1.9 KiB
 python-srpm-macros                noarch 0:3.14-11.fc45               fedora      51.6 KiB
 qt5-srpm-macros                   noarch 0:5.15.18-2.fc44             fedora     500.0   B
 qt6-srpm-macros                   noarch 0:6.10.2-1.fc45              fedora     472.0   B
 readline                          x86_64 0:8.3-4.fc44                 fedora     519.5 KiB
 rpm                               x86_64 0:6.0.1-5.fc45               fedora       3.1 MiB
 rpm-build-libs                    x86_64 0:6.0.1-5.fc45               fedora     276.3 KiB
 rpm-libs                          x86_64 0:6.0.1-5.fc45               fedora     961.2 KiB
 rpm-plugin-selinux                x86_64 0:6.0.1-5.fc45               fedora      11.9 KiB
 rpm-sequoia                       x86_64 0:1.10.1-1.fc45              fedora       2.4 MiB
 rpm-sign-libs                     x86_64 0:6.0.1-5.fc45               fedora      39.6 KiB
 rust-srpm-macros                  noarch 0:28.4-3.fc44                fedora       5.5 KiB
 selinux-policy                    noarch 0:43.1-1.fc45                fedora      32.0 KiB
 selinux-policy-targeted           noarch 0:43.1-1.fc45                fedora      18.5 MiB
 setup                             noarch 0:2.15.0-29.fc45             fedora     724.9 KiB
 sqlite-libs                       x86_64 0:3.52.0-1.fc45              fedora       1.6 MiB
 systemd-libs                      x86_64 0:260~rc4-1.fc45             fedora       2.5 MiB
 systemd-standalone-sysusers       x86_64 0:260~rc4-1.fc45             fedora     841.2 KiB
 tpm2-tss                          x86_64 0:4.1.3-9.fc44               fedora       1.6 MiB
 tree-sitter-srpm-macros           noarch 0:0.4.2-2.fc44               fedora       8.3 KiB
 util-linux-core                   x86_64 0:2.41.3-12.fc44             fedora       1.5 MiB
 xxhash-libs                       x86_64 0:0.8.3-4.fc44               fedora      94.0 KiB
 xz-libs                           x86_64 1:5.8.2-2.fc44               fedora     217.7 KiB
 zig-srpm-macros                   noarch 0:1-8.fc44                   fedora       1.3 KiB
 zip                               x86_64 0:3.0-45.fc44                fedora     698.0 KiB
 zlib-ng-compat                    x86_64 0:2.3.3-5.fc45               fedora     165.6 KiB
 zstd                              x86_64 0:1.5.7-5.fc44               fedora     502.4 KiB
Installing groups:
 Buildsystem building group                                                                

Transaction Summary:
 Installing:       185 packages

Total size of inbound packages is 70 MiB. Need to download 70 MiB.
After this operation, 226 MiB extra will be used (install 226 MiB, remove 0 B).
[  1/185] bzip2-0:1.0.8-23.fc44.x86_64  100% |   3.6 MiB/s |  52.1 KiB |  00m00s
[  2/185] bash-0:5.3.9-3.fc44.x86_64    100% | 105.2 MiB/s |   1.9 MiB |  00m00s
[  3/185] cpio-0:2.15-9.fc44.x86_64     100% |  57.3 MiB/s | 293.4 KiB |  00m00s
[  4/185] diffutils-0:3.12-5.fc44.x86_6 100% |  77.2 MiB/s | 395.3 KiB |  00m00s
[  5/185] coreutils-0:9.10-3.fc45.x86_6 100% |  47.6 MiB/s |   1.2 MiB |  00m00s
[  6/185] fedora-release-common-0:45-0. 100% |   4.7 MiB/s |  23.9 KiB |  00m00s
[  7/185] findutils-1:4.10.0-7.fc44.x86 100% |  77.1 MiB/s | 552.5 KiB |  00m00s
[  8/185] glibc-minimal-langpack-0:2.43 100% |  13.9 MiB/s |  85.4 KiB |  00m00s
[  9/185] grep-0:3.12-3.fc44.x86_64     100% |  29.4 MiB/s | 301.6 KiB |  00m00s
[ 10/185] gzip-0:1.14-2.fc44.x86_64     100% |  25.0 MiB/s | 179.1 KiB |  00m00s
[ 11/185] info-0:7.3-1.fc45.x86_64      100% |  18.6 MiB/s | 190.8 KiB |  00m00s
[ 12/185] patch-0:2.8-4.fc44.x86_64     100% |  18.8 MiB/s | 115.2 KiB |  00m00s
[ 13/185] redhat-rpm-config-0:344-1.fc4 100% |   8.5 MiB/s |  78.4 KiB |  00m00s
[ 14/185] rpm-build-0:6.0.1-5.fc45.x86_ 100% |  19.5 MiB/s | 139.9 KiB |  00m00s
[ 15/185] sed-0:4.9-8.fc45.x86_64       100% |  28.3 MiB/s | 319.0 KiB |  00m00s
[ 16/185] shadow-utils-2:4.19.3-1.fc45. 100% | 144.1 MiB/s |   1.3 MiB |  00m00s
[ 17/185] tar-2:1.35-8.fc44.x86_64      100% |  49.9 MiB/s | 868.2 KiB |  00m00s
[ 18/185] unzip-0:6.0-69.fc44.x86_64    100% |  15.0 MiB/s | 199.6 KiB |  00m00s
[ 19/185] which-0:2.23-4.fc44.x86_64    100% |   4.6 MiB/s |  42.2 KiB |  00m00s
[ 20/185] xz-1:5.8.2-2.fc44.x86_64      100% |  53.9 MiB/s | 607.2 KiB |  00m00s
[ 21/185] gawk-0:5.4.0-2.fc45.x86_64    100% |  61.5 MiB/s |   1.2 MiB |  00m00s
[ 22/185] util-linux-0:2.41.3-12.fc44.x 100% |  60.2 MiB/s |   1.2 MiB |  00m00s
[ 23/185] filesystem-0:3.18-56.fc45.x86 100% |  87.3 MiB/s |   1.8 MiB |  00m00s
[ 24/185] glibc-0:2.43.9000-4.fc45.x86_ 100% | 125.1 MiB/s |   2.4 MiB |  00m00s
[ 25/185] ncurses-libs-0:6.6-1.fc44.x86 100% |  18.7 MiB/s | 344.1 KiB |  00m00s
[ 26/185] bzip2-libs-0:1.0.8-23.fc44.x8 100% |   6.1 MiB/s |  43.8 KiB |  00m00s
[ 27/185] coreutils-common-0:9.10-3.fc4 100% | 237.0 MiB/s |   2.1 MiB |  00m00s
[ 28/185] gmp-1:6.3.0-5.fc44.x86_64     100% |  35.0 MiB/s | 322.5 KiB |  00m00s
[ 29/185] libacl-0:2.3.2-6.fc44.x86_64  100% |   2.7 MiB/s |  24.8 KiB |  00m00s
[ 30/185] libattr-0:2.5.2-8.fc44.x86_64 100% |   8.8 MiB/s |  17.9 KiB |  00m00s
[ 31/185] libcap-0:2.77-2.fc44.x86_64   100% |  28.8 MiB/s |  88.5 KiB |  00m00s
[ 32/185] libselinux-0:3.10-1.fc44.x86_ 100% |  24.9 MiB/s | 102.0 KiB |  00m00s
[ 33/185] openssl-libs-1:3.5.5-1.fc44.x 100% | 171.5 MiB/s |   2.7 MiB |  00m00s
[ 34/185] systemd-libs-0:260~rc4-1.fc45 100% |  54.1 MiB/s | 887.0 KiB |  00m00s
[ 35/185] fedora-repos-0:45-0.1.noarch  100% | 542.6 KiB/s |   9.2 KiB |  00m00s
[ 36/185] glibc-common-0:2.43.9000-4.fc 100% |  46.2 MiB/s | 378.7 KiB |  00m00s
[ 37/185] pcre2-0:10.47-1.fc44.1.x86_64 100% |  30.1 MiB/s | 277.7 KiB |  00m00s
[ 38/185] ed-0:1.22.5-2.fc45.x86_64     100% |   8.3 MiB/s |  85.4 KiB |  00m00s
[ 39/185] R-srpm-macros-0:1.3.5-1.fc45. 100% |   2.1 MiB/s |  10.8 KiB |  00m00s
[ 40/185] ansible-srpm-macros-0:1-20.1. 100% |   4.9 MiB/s |  20.1 KiB |  00m00s
[ 41/185] build-reproducibility-srpm-ma 100% |   3.2 MiB/s |  13.3 KiB |  00m00s
[ 42/185] cmake-srpm-macros-0:4.2.3-2.f 100% |   2.6 MiB/s |  10.5 KiB |  00m00s
[ 43/185] dwz-0:0.16-3.fc44.x86_64      100% |  33.9 MiB/s | 138.9 KiB |  00m00s
[ 44/185] efi-srpm-macros-0:6-6.fc44.no 100% |   7.3 MiB/s |  22.6 KiB |  00m00s
[ 45/185] erlang-srpm-macros-0:0.3.11-1 100% |   2.4 MiB/s |  10.0 KiB |  00m00s
[ 46/185] file-0:5.47-1.fc45.x86_64     100% |  12.0 MiB/s |  49.1 KiB |  00m00s
[ 47/185] filesystem-srpm-macros-0:3.18 100% |  13.1 MiB/s |  26.8 KiB |  00m00s
[ 48/185] fonts-srpm-macros-1:5.0.0-3.f 100% |  13.3 MiB/s |  27.2 KiB |  00m00s
[ 49/185] forge-srpm-macros-0:0.4.0-4.f 100% |   9.8 MiB/s |  20.0 KiB |  00m00s
[ 50/185] fpc-srpm-macros-0:1.3-16.fc44 100% |   3.9 MiB/s |   7.9 KiB |  00m00s
[ 51/185] gap-srpm-macros-0:2-2.fc44.no 100% |   2.2 MiB/s |   9.1 KiB |  00m00s
[ 52/185] ghc-srpm-macros-0:1.10-1.fc44 100% |   2.1 MiB/s |   8.8 KiB |  00m00s
[ 53/185] gnat-srpm-macros-0:7-2.fc44.n 100% |   2.8 MiB/s |   8.7 KiB |  00m00s
[ 54/185] go-srpm-macros-0:3.8.0-2.fc44 100% |   9.2 MiB/s |  28.2 KiB |  00m00s
[ 55/185] java-srpm-macros-0:1-8.fc44.n 100% |   2.6 MiB/s |   8.1 KiB |  00m00s
[ 56/185] kernel-srpm-macros-0:1.0-28.f 100% |   4.4 MiB/s |   9.0 KiB |  00m00s
[ 57/185] lua-srpm-macros-0:1-17.fc44.n 100% |   4.3 MiB/s |   8.9 KiB |  00m00s
[ 58/185] ocaml-srpm-macros-0:11-3.fc44 100% |   4.5 MiB/s |   9.3 KiB |  00m00s
[ 59/185] openblas-srpm-macros-0:2-21.f 100% |   3.8 MiB/s |   7.8 KiB |  00m00s
[ 60/185] package-notes-srpm-macros-0:0 100% |   4.7 MiB/s |   9.7 KiB |  00m00s
[ 61/185] perl-srpm-macros-0:1-61.fc44. 100% |   4.1 MiB/s |   8.4 KiB |  00m00s
[ 62/185] pyproject-srpm-macros-0:1.18. 100% |   6.4 MiB/s |  13.1 KiB |  00m00s
[ 63/185] python-srpm-macros-0:3.14-11. 100% |  11.6 MiB/s |  23.8 KiB |  00m00s
[ 64/185] qt5-srpm-macros-0:5.15.18-2.f 100% |   4.2 MiB/s |   8.7 KiB |  00m00s
[ 65/185] qt6-srpm-macros-0:6.10.2-1.fc 100% |   4.4 MiB/s |   9.1 KiB |  00m00s
[ 66/185] rpm-0:6.0.1-5.fc45.x86_64     100% | 141.2 MiB/s | 578.3 KiB |  00m00s
[ 67/185] rust-srpm-macros-0:28.4-3.fc4 100% |   3.6 MiB/s |  11.1 KiB |  00m00s
[ 68/185] tree-sitter-srpm-macros-0:0.4 100% |   4.4 MiB/s |  13.5 KiB |  00m00s
[ 69/185] zig-srpm-macros-0:1-8.fc44.no 100% |   4.3 MiB/s |   8.8 KiB |  00m00s
[ 70/185] zip-0:3.0-45.fc44.x86_64      100% | 129.1 MiB/s | 264.4 KiB |  00m00s
[ 71/185] debugedit-0:5.3-2.fc45.x86_64 100% |  28.8 MiB/s |  88.5 KiB |  00m00s
[ 72/185] elfutils-0:0.194-5.fc45.x86_6 100% | 142.7 MiB/s | 584.4 KiB |  00m00s
[ 73/185] elfutils-libelf-0:0.194-5.fc4 100% |  50.1 MiB/s | 205.1 KiB |  00m00s
[ 74/185] libarchive-0:3.8.6-1.fc45.x86 100% |  86.4 MiB/s | 442.6 KiB |  00m00s
[ 75/185] popt-0:1.19-10.fc44.x86_64    100% |  13.1 MiB/s |  66.9 KiB |  00m00s
[ 76/185] readline-0:8.3-4.fc44.x86_64  100% |  56.1 MiB/s | 229.8 KiB |  00m00s
[ 77/185] rpm-build-libs-0:6.0.1-5.fc45 100% |  25.8 MiB/s | 132.1 KiB |  00m00s
[ 78/185] rpm-libs-0:6.0.1-5.fc45.x86_6 100% | 101.4 MiB/s | 415.5 KiB |  00m00s
[ 79/185] zstd-0:1.5.7-5.fc44.x86_64    100% |  47.0 MiB/s | 192.5 KiB |  00m00s
[ 80/185] audit-libs-0:4.1.3-1.fc44.x86 100% |  46.3 MiB/s | 142.1 KiB |  00m00s
[ 81/185] libeconf-0:0.7.9-3.fc44.x86_6 100% |  17.7 MiB/s |  36.2 KiB |  00m00s
[ 82/185] libsemanage-0:3.10-1.fc44.x86 100% |  62.1 MiB/s | 127.3 KiB |  00m00s
[ 83/185] libxcrypt-0:4.5.2-3.fc44.x86_ 100% |  64.1 MiB/s | 131.2 KiB |  00m00s
[ 84/185] pam-libs-0:1.7.2-1.fc44.x86_6 100% |  14.4 MiB/s |  59.1 KiB |  00m00s
[ 85/185] setup-0:2.15.0-29.fc45.noarch 100% |  51.2 MiB/s | 157.3 KiB |  00m00s
[ 86/185] xz-libs-1:5.8.2-2.fc44.x86_64 100% |  38.0 MiB/s | 116.8 KiB |  00m00s
[ 87/185] mpfr-0:4.2.2-3.fc44.x86_64    100% | 116.5 MiB/s | 357.8 KiB |  00m00s
[ 88/185] libblkid-0:2.41.3-12.fc44.x86 100% |  42.1 MiB/s | 129.4 KiB |  00m00s
[ 89/185] libcap-ng-0:0.9.1-1.fc45.x86_ 100% |  16.0 MiB/s |  32.8 KiB |  00m00s
[ 90/185] libfdisk-0:2.41.3-12.fc44.x86 100% |  82.1 MiB/s | 168.1 KiB |  00m00s
[ 91/185] liblastlog2-0:2.41.3-12.fc44. 100% |   7.6 MiB/s |  23.5 KiB |  00m00s
[ 92/185] libmount-0:2.41.3-12.fc44.x86 100% |  41.7 MiB/s | 171.0 KiB |  00m00s
[ 93/185] libsmartcols-0:2.41.3-12.fc44 100% |  28.3 MiB/s |  86.9 KiB |  00m00s
[ 94/185] libuuid-0:2.41.3-12.fc44.x86_ 100% |   8.7 MiB/s |  26.6 KiB |  00m00s
[ 95/185] util-linux-core-0:2.41.3-12.f 100% | 136.2 MiB/s | 557.8 KiB |  00m00s
[ 96/185] zlib-ng-compat-0:2.3.3-5.fc45 100% |  29.7 MiB/s |  91.2 KiB |  00m00s
[ 97/185] glibc-gconv-extra-0:2.43.9000 100% | 184.3 MiB/s |   1.7 MiB |  00m00s
[ 98/185] ncurses-base-0:6.6-1.fc44.noa 100% |  12.3 MiB/s |  88.0 KiB |  00m00s
[ 99/185] gnulib-l10n-0:20241231-2.fc44 100% |  24.5 MiB/s | 150.3 KiB |  00m00s
[100/185] libsepol-0:3.10-1.fc44.x86_64 100% |  89.8 MiB/s | 367.6 KiB |  00m00s
[101/185] ca-certificates-0:2025.2.80_v 100% | 135.7 MiB/s | 972.8 KiB |  00m00s
[102/185] crypto-policies-0:20251128-3. 100% |  13.7 MiB/s |  98.3 KiB |  00m00s
[103/185] fedora-gpg-keys-0:45-0.1.noar 100% |  34.5 MiB/s | 141.3 KiB |  00m00s
[104/185] fedora-repos-rawhide-0:45-0.1 100% |   4.3 MiB/s |   8.8 KiB |  00m00s
[105/185] pcre2-syntax-0:10.47-1.fc44.1 100% |  53.6 MiB/s | 164.7 KiB |  00m00s
[106/185] add-determinism-0:0.7.3-1.fc4 100% | 119.2 MiB/s | 854.5 KiB |  00m00s
[107/185] linkdupes-0:0.7.3-1.fc45.x86_ 100% |  56.1 MiB/s | 344.5 KiB |  00m00s
[108/185] file-libs-0:5.47-1.fc45.x86_6 100% | 106.4 MiB/s | 871.9 KiB |  00m00s
[109/185] curl-0:8.19.0~rc3-1.fc45.x86_ 100% |  47.2 MiB/s | 241.9 KiB |  00m00s
[110/185] elfutils-libs-0:0.194-5.fc45. 100% |  46.1 MiB/s | 283.0 KiB |  00m00s
[111/185] elfutils-debuginfod-client-0: 100% |  15.2 MiB/s |  46.7 KiB |  00m00s
[112/185] libzstd-0:1.5.7-5.fc44.x86_64 100% | 118.5 MiB/s | 364.2 KiB |  00m00s
[113/185] libxml2-0:2.12.10-6.fc44.x86_ 100% | 176.9 MiB/s | 724.6 KiB |  00m00s
[114/185] lz4-libs-0:1.10.0-4.fc44.x86_ 100% |  19.0 MiB/s |  77.7 KiB |  00m00s
[115/185] lua-libs-0:5.5.0-1.fc45.x86_6 100% |  35.1 MiB/s | 143.6 KiB |  00m00s
[116/185] rpm-sign-libs-0:6.0.1-5.fc45. 100% |  13.6 MiB/s |  27.9 KiB |  00m00s
[117/185] rpm-sequoia-0:1.10.1-1.fc45.x 100% | 178.6 MiB/s | 914.2 KiB |  00m00s
[118/185] sqlite-libs-0:3.52.0-1.fc45.x 100% |  78.4 MiB/s | 802.7 KiB |  00m00s
[119/185] elfutils-default-yama-scope-0 100% |   1.3 MiB/s |  11.9 KiB |  00m00s
[120/185] json-c-0:0.18-8.fc44.x86_64   100% |   6.5 MiB/s |  46.3 KiB |  00m00s
[121/185] gnupg2-0:2.4.9-5.fc44.x86_64  100% | 236.1 MiB/s |   1.7 MiB |  00m00s
[122/185] ima-evm-utils-libs-0:1.6.2-10 100% |   3.7 MiB/s |  30.4 KiB |  00m00s
[123/185] libfsverity-0:1.7-1.fc45.x86_ 100% |   2.7 MiB/s |  19.0 KiB |  00m00s
[124/185] gpgverify-0:2.2-4.fc44.noarch 100% |   5.5 MiB/s |  11.2 KiB |  00m00s
[125/185] gnupg2-dirmngr-0:2.4.9-5.fc44 100% |  91.9 MiB/s | 282.4 KiB |  00m00s
[126/185] gnupg2-gpg-agent-0:2.4.9-5.fc 100% |  45.8 MiB/s | 281.3 KiB |  00m00s
[127/185] gnupg2-gpgconf-0:2.4.9-5.fc44 100% |  19.1 MiB/s | 117.1 KiB |  00m00s
[128/185] gnupg2-keyboxd-0:2.4.9-5.fc44 100% |  15.8 MiB/s |  97.3 KiB |  00m00s
[129/185] gnupg2-verify-0:2.4.9-5.fc44. 100% |  43.2 MiB/s | 176.8 KiB |  00m00s
[130/185] libassuan-0:2.5.7-5.fc44.x86_ 100% |   9.6 MiB/s |  68.9 KiB |  00m00s
[131/185] libgcrypt-0:1.12.1-1.fc45.x86 100% |  55.9 MiB/s | 629.4 KiB |  00m00s
[132/185] libgpg-error-0:1.59-1.fc45.x8 100% |  27.5 MiB/s | 253.8 KiB |  00m00s
[133/185] npth-0:1.8-4.fc44.x86_64      100% |   3.1 MiB/s |  25.8 KiB |  00m00s
[134/185] tpm2-tss-0:4.1.3-9.fc44.x86_6 100% |  71.8 MiB/s | 440.9 KiB |  00m00s
[135/185] gnutls-0:3.8.12-1.fc45.x86_64 100% | 107.9 MiB/s |   1.4 MiB |  00m00s
[136/185] libksba-0:1.6.8-1.fc45.x86_64 100% |  13.7 MiB/s | 168.9 KiB |  00m00s
[137/185] openldap-0:2.6.13-1.fc45.x86_ 100% |  24.1 MiB/s | 272.0 KiB |  00m00s
[138/185] libusb1-0:1.0.29-5.fc44.x86_6 100% |  13.5 MiB/s |  82.8 KiB |  00m00s
[139/185] libidn2-0:2.3.8-3.fc44.x86_64 100% |  24.5 MiB/s | 175.5 KiB |  00m00s
[140/185] libtasn1-0:4.21.0-1.fc45.x86_ 100% |  12.4 MiB/s |  76.0 KiB |  00m00s
[141/185] libunistring-0:1.1-11.fc44.x8 100% |  89.2 MiB/s | 547.8 KiB |  00m00s
[142/185] nettle-0:3.10.1-3.fc44.x86_64 100% |  69.7 MiB/s | 428.0 KiB |  00m00s
[143/185] p11-kit-0:0.26.2-1.fc45.x86_6 100% |  90.5 MiB/s | 555.9 KiB |  00m00s
[144/185] cyrus-sasl-lib-0:2.1.28-35.fc 100% |  97.7 MiB/s | 800.7 KiB |  00m00s
[145/185] libevent-0:2.1.12-17.fc44.x86 100% |  40.1 MiB/s | 287.4 KiB |  00m00s
[146/185] libtool-ltdl-0:2.5.4-10.fc44. 100% |   7.2 MiB/s |  37.0 KiB |  00m00s
[147/185] libffi-0:3.5.2-2.fc44.x86_64  100% |  41.4 MiB/s |  42.4 KiB |  00m00s
[148/185] gdbm-libs-1:1.23-11.fc44.x86_ 100% |  57.4 MiB/s |  58.8 KiB |  00m00s
[149/185] libgcc-0:16.0.1-0.9.fc45.x86_ 100% |  10.1 MiB/s | 123.7 KiB |  00m00s
[150/185] libgomp-0:16.0.1-0.9.fc45.x86 100% |  28.9 MiB/s | 384.4 KiB |  00m00s
[151/185] libstdc++-0:16.0.1-0.9.fc45.x 100% |  57.7 MiB/s | 945.9 KiB |  00m00s
[152/185] alternatives-0:1.33-5.fc44.x8 100% |  19.9 MiB/s |  40.8 KiB |  00m00s
[153/185] jansson-0:2.14-4.fc44.x86_64  100% |  23.0 MiB/s |  47.1 KiB |  00m00s
[154/185] pkgconf-pkg-config-0:2.5.1-1. 100% |   2.3 MiB/s |   9.5 KiB |  00m00s
[155/185] pkgconf-0:2.5.1-1.fc45.x86_64 100% |  11.9 MiB/s |  48.7 KiB |  00m00s
[156/185] pkgconf-m4-0:2.5.1-1.fc45.noa 100% |   6.7 MiB/s |  13.8 KiB |  00m00s
[157/185] libpkgconf-0:2.5.1-1.fc45.x86 100% |  20.8 MiB/s |  42.7 KiB |  00m00s
[158/185] p11-kit-trust-0:0.26.2-1.fc45 100% |  71.5 MiB/s | 146.4 KiB |  00m00s
[159/185] fedora-release-0:45-0.2.noarc 100% |   6.2 MiB/s |  12.8 KiB |  00m00s
[160/185] systemd-standalone-sysusers-0 100% | 119.7 MiB/s | 367.7 KiB |  00m00s
[161/185] gdb-minimal-0:17.1-5.fc45.x86 100% | 189.4 MiB/s |   4.7 MiB |  00m00s
[162/185] xxhash-libs-0:0.8.3-4.fc44.x8 100% |   1.8 MiB/s |  41.2 KiB |  00m00s
[163/185] fedora-release-identity-basic 100% |   6.6 MiB/s |  13.5 KiB |  00m00s
[164/185] libcurl-0:8.19.0~rc3-1.fc45.x 100% | 111.5 MiB/s | 456.6 KiB |  00m00s
[165/185] krb5-libs-0:1.22.2-2.fc45.x86 100% | 156.9 MiB/s | 803.3 KiB |  00m00s
[166/185] libbrotli-0:1.2.0-3.fc44.x86_ 100% |  84.9 MiB/s | 347.8 KiB |  00m00s
[167/185] libnghttp2-0:1.68.0-3.fc44.x8 100% |  37.7 MiB/s |  77.2 KiB |  00m00s
[168/185] libnghttp3-0:1.15.0-1.fc44.x8 100% |  36.1 MiB/s |  73.9 KiB |  00m00s
[169/185] libpsl-0:0.21.5-7.fc44.x86_64 100% |  31.9 MiB/s |  65.3 KiB |  00m00s
[170/185] libssh-0:0.12.0-1.fc45.x86_64 100% |  95.6 MiB/s | 293.7 KiB |  00m00s
[171/185] ngtcp2-0:1.19.0-2.fc44.x86_64 100% |  50.4 MiB/s | 154.8 KiB |  00m00s
[172/185] ngtcp2-crypto-ossl-0:1.19.0-2 100% |  13.4 MiB/s |  27.4 KiB |  00m00s
[173/185] keyutils-libs-0:1.6.3-7.fc44. 100% |  15.4 MiB/s |  31.5 KiB |  00m00s
[174/185] libcom_err-0:1.47.4-1.fc45.x8 100% |   8.7 MiB/s |  26.6 KiB |  00m00s
[175/185] libverto-0:0.3.2-12.fc44.x86_ 100% |   5.2 MiB/s |  21.1 KiB |  00m00s
[176/185] publicsuffix-list-dafsa-0:202 100% |  19.6 MiB/s |  60.3 KiB |  00m00s
[177/185] libfido2-0:1.16.0-5.fc44.x86_ 100% |  34.0 MiB/s | 104.5 KiB |  00m00s
[178/185] libssh-config-0:0.12.0-1.fc45 100% |   4.6 MiB/s |   9.4 KiB |  00m00s
[179/185] libcbor-0:0.13.0-2.fc44.x86_6 100% |  17.6 MiB/s |  36.0 KiB |  00m00s
[180/185] policycoreutils-0:3.10-2.fc45 100% |  14.9 MiB/s | 258.8 KiB |  00m00s
[181/185] selinux-policy-targeted-0:43. 100% | 235.1 MiB/s |   6.8 MiB |  00m00s
[182/185] selinux-policy-0:43.1-1.fc45. 100% |   6.3 MiB/s |  70.4 KiB |  00m00s
[183/185] libselinux-utils-0:3.10-1.fc4 100% |  58.4 MiB/s | 119.6 KiB |  00m00s
[184/185] rpm-plugin-selinux-0:6.0.1-5. 100% |   6.2 MiB/s |  19.1 KiB |  00m00s
[185/185] binutils-0:2.46.50-3.fc45.x86 100% |   9.5 MiB/s |   6.2 MiB |  00m01s
--------------------------------------------------------------------------------
[185/185] Total                         100% |  69.3 MiB/s |  69.7 MiB |  00m01s
Running transaction
Importing OpenPGP key 0xF577861E:
 UserID     : "Fedora (45) <fedora-45-primary@fedoraproject.org>"
 Fingerprint: 4F50A6114CD5C6976A7F1179655A4B02F577861E
 From       : file:///usr/share/distribution-gpg-keys/fedora/RPM-GPG-KEY-fedora-45-primary
The key was successfully imported.
Importing OpenPGP key 0xF577861E:
 UserID     : "Fedora (45) <fedora-45-primary@fedoraproject.org>"
 Fingerprint: 4F50A6114CD5C6976A7F1179655A4B02F577861E
 From       : file:///usr/share/distribution-gpg-keys/fedora/RPM-GPG-KEY-fedora-45-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.
Importing OpenPGP key 0x91211FCE:
 UserID     : "Fedora (46) <fedora-46-primary@fedoraproject.org>"
 Fingerprint: D924B10D3E810DABDD8B56B596E7E91491211FCE
 From       : file:///usr/share/distribution-gpg-keys/fedora/RPM-GPG-KEY-fedora-46-primary
The key was successfully imported.
[  1/187] Verify package files          100% | 627.0   B/s | 185.0   B |  00m00s
>>> Running %pretrans scriptlet: filesystem-0:3.18-56.fc45.x86_64
>>> Finished %pretrans scriptlet: filesystem-0:3.18-56.fc45.x86_64
>>> [RPM] /var/lib/mock/fedora-rawhide-x86_64-1773846985.598827/root/var/cache/dnf/copr_base-6729cd57cc4ff6a8/packages/libgcc-16.0.1-0.9.fc45.x86_64.rpm: Header OpenPGP V4 RSA/SHA256 signature, key ID 4d0ea48a8d983303: NOKEY
[  2/187] Prepare transaction           100% |   2.9 KiB/s | 185.0   B |  00m00s
[  3/187] Installing libgcc-0:16.0.1-0. 100% | 266.0 MiB/s | 272.4 KiB |  00m00s
[  4/187] Installing libssh-config-0:0. 100% |   0.0   B/s | 816.0   B |  00m00s
[  5/187] Installing publicsuffix-list- 100% |   0.0   B/s |  71.1 KiB |  00m00s
[  6/187] Installing fedora-release-ide 100% |   0.0   B/s | 920.0   B |  00m00s
[  7/187] Installing fedora-gpg-keys-0: 100% |  44.5 MiB/s | 182.1 KiB |  00m00s
[  8/187] Installing fedora-repos-rawhi 100% |   0.0   B/s |   2.4 KiB |  00m00s
[  9/187] Installing fedora-repos-0:45- 100% |   0.0   B/s |   5.7 KiB |  00m00s
[ 10/187] Installing fedora-release-com 100% |  24.3 MiB/s |  24.9 KiB |  00m00s
[ 11/187] Installing fedora-release-0:4 100% |  15.1 KiB/s | 124.0   B |  00m00s
>>> Running sysusers scriptlet: setup-0:2.15.0-29.fc45.noarch
>>> Finished sysusers scriptlet: setup-0:2.15.0-29.fc45.noarch
>>> Scriptlet output:
>>> Creating group 'adm' with GID 4.
>>> Creating group 'audio' with GID 63.
>>> Creating group 'cdrom' with GID 11.
>>> Creating group 'clock' with GID 103.
>>> Creating group 'dialout' with GID 18.
>>> Creating group 'disk' with GID 6.
>>> Creating group 'floppy' with GID 19.
>>> Creating group 'ftp' with GID 50.
>>> Creating group 'games' with GID 20.
>>> Creating group 'input' with GID 104.
>>> Creating group 'kmem' with GID 9.
>>> Creating group 'kvm' with GID 36.
>>> Creating group 'lock' with GID 54.
>>> Creating group 'lp' with GID 7.
>>> Creating group 'mail' with GID 12.
>>> Creating group 'man' with GID 15.
>>> Creating group 'mem' with GID 8.
>>> Creating group 'nobody' with GID 65534.
>>> Creating group 'render' with GID 105.
>>> Creating group 'root' with GID 0.
>>> Creating group 'sgx' with GID 106.
>>> Creating group 'sys' with GID 3.
>>> Creating group 'tape' with GID 33.
>>> Creating group 'tty' with GID 5.
>>> Creating group 'users' with GID 100.
>>> Creating group 'utmp' with GID 22.
>>> Creating group 'video' with GID 39.
>>> Creating group 'wheel' with GID 10.
>>> Creating user 'adm' (adm) with UID 3 and GID 4.
>>> Creating group 'bin' with GID 1.
>>> Creating user 'bin' (bin) with UID 1 and GID 1.
>>> Creating group 'daemon' with GID 2.
>>> Creating user 'daemon' (daemon) with UID 2 and GID 2.
>>> Creating user 'ftp' (FTP User) with UID 14 and GID 50.
>>> Creating user 'games' (games) with UID 12 and GID 100.
>>> Creating user 'halt' (halt) with UID 7 and GID 0.
>>> Creating user 'lp' (lp) with UID 4 and GID 7.
>>> Creating user 'mail' (mail) with UID 8 and GID 12.
>>> Creating user 'nobody' (Kernel Overflow User) with UID 65534 and GID 65534.
>>> Creating user 'operator' (operator) with UID 11 and GID 0.
>>> Creating user 'root' (Super User) with UID 0 and GID 0.
>>> Creating user 'shutdown' (shutdown) with UID 6 and GID 0.
>>> Creating user 'sync' (sync) with UID 5 and GID 0.
>>> 
[ 12/187] Installing setup-0:2.15.0-29. 100% |  35.7 MiB/s | 730.6 KiB |  00m00s
>>> [RPM] /etc/hosts created as /etc/hosts.rpmnew
[ 13/187] Installing filesystem-0:3.18- 100% |   2.5 MiB/s | 289.4 KiB |  00m00s
[ 14/187] Installing pkgconf-m4-0:2.5.1 100% |   0.0   B/s |  14.7 KiB |  00m00s
[ 15/187] Installing pcre2-syntax-0:10. 100% | 277.7 MiB/s | 284.3 KiB |  00m00s
[ 16/187] Installing gnulib-l10n-0:2024 100% | 215.5 MiB/s | 661.9 KiB |  00m00s
[ 17/187] Installing coreutils-common-0 100% | 356.2 MiB/s |  10.7 MiB |  00m00s
[ 18/187] Installing ncurses-base-0:6.6 100% |  86.7 MiB/s | 355.3 KiB |  00m00s
[ 19/187] Installing bash-0:5.3.9-3.fc4 100% | 223.2 MiB/s |   8.5 MiB |  00m00s
[ 20/187] Installing glibc-common-0:2.4 100% |  44.8 MiB/s |   1.0 MiB |  00m00s
[ 21/187] Installing glibc-gconv-extra- 100% | 277.1 MiB/s |   7.8 MiB |  00m00s
[ 22/187] Installing glibc-0:2.43.9000- 100% | 158.3 MiB/s |   7.0 MiB |  00m00s
[ 23/187] Installing ncurses-libs-0:6.6 100% | 238.1 MiB/s | 975.4 KiB |  00m00s
[ 24/187] Installing glibc-minimal-lang 100% |   0.0   B/s | 124.0   B |  00m00s
[ 25/187] Installing zlib-ng-compat-0:2 100% | 162.5 MiB/s | 166.4 KiB |  00m00s
[ 26/187] Installing bzip2-libs-0:1.0.8 100% |  79.7 MiB/s |  81.6 KiB |  00m00s
[ 27/187] Installing libgpg-error-0:1.5 100% |  44.3 MiB/s | 952.3 KiB |  00m00s
[ 28/187] Installing libstdc++-0:16.0.1 100% | 365.3 MiB/s |   2.9 MiB |  00m00s
[ 29/187] Installing libassuan-0:2.5.7- 100% | 161.8 MiB/s | 165.7 KiB |  00m00s
[ 30/187] Installing libgcrypt-0:1.12.1 100% | 331.6 MiB/s |   1.7 MiB |  00m00s
[ 31/187] Installing readline-0:8.3-4.f 100% | 254.7 MiB/s | 521.6 KiB |  00m00s
[ 32/187] Installing gmp-1:6.3.0-5.fc44 100% | 266.1 MiB/s | 817.5 KiB |  00m00s
[ 33/187] Installing xz-libs-1:5.8.2-2. 100% | 213.7 MiB/s | 218.8 KiB |  00m00s
[ 34/187] Installing libuuid-0:2.41.3-1 100% |  37.3 MiB/s |  38.2 KiB |  00m00s
[ 35/187] Installing systemd-libs-0:260 100% | 363.9 MiB/s |   2.5 MiB |  00m00s
[ 36/187] Installing popt-0:1.19-10.fc4 100% |  68.0 MiB/s | 139.3 KiB |  00m00s
[ 37/187] Installing libzstd-0:1.5.7-5. 100% | 311.6 MiB/s | 957.4 KiB |  00m00s
[ 38/187] Installing elfutils-libelf-0: 100% | 376.3 MiB/s |   1.1 MiB |  00m00s
[ 39/187] Installing npth-0:1.8-4.fc44. 100% |   0.0   B/s |  50.6 KiB |  00m00s
[ 40/187] Installing libblkid-0:2.41.3- 100% | 269.0 MiB/s | 275.4 KiB |  00m00s
[ 41/187] Installing libsepol-0:3.10-1. 100% | 283.5 MiB/s | 871.0 KiB |  00m00s
[ 42/187] Installing sqlite-libs-0:3.52 100% | 320.6 MiB/s |   1.6 MiB |  00m00s
[ 43/187] Installing gnupg2-gpgconf-0:2 100% |  12.9 MiB/s | 251.8 KiB |  00m00s
[ 44/187] Installing libattr-0:2.5.2-8. 100% |  24.7 MiB/s |  25.2 KiB |  00m00s
[ 45/187] Installing libacl-0:2.3.2-6.f 100% |   0.0   B/s |  36.6 KiB |  00m00s
[ 46/187] Installing pcre2-0:10.47-1.fc 100% | 351.6 MiB/s | 720.0 KiB |  00m00s
[ 47/187] Installing libselinux-0:3.10- 100% | 197.5 MiB/s | 202.3 KiB |  00m00s
[ 48/187] Installing grep-0:3.12-3.fc44 100% |  45.6 MiB/s |   1.0 MiB |  00m00s
[ 49/187] Installing sed-0:4.9-8.fc45.x 100% |  40.2 MiB/s | 865.4 KiB |  00m00s
[ 50/187] Installing findutils-1:4.10.0 100% |  81.3 MiB/s |   1.9 MiB |  00m00s
[ 51/187] Installing libxcrypt-0:4.5.2- 100% | 144.5 MiB/s | 295.9 KiB |  00m00s
[ 52/187] Installing libtasn1-0:4.21.0- 100% | 178.2 MiB/s | 182.4 KiB |  00m00s
[ 53/187] Installing libunistring-0:1.1 100% | 346.9 MiB/s |   1.7 MiB |  00m00s
[ 54/187] Installing libidn2-0:2.3.8-3. 100% |  61.0 MiB/s | 562.6 KiB |  00m00s
[ 55/187] Installing crypto-policies-0: 100% |  30.8 MiB/s | 157.7 KiB |  00m00s
[ 56/187] Installing xz-1:5.8.2-2.fc44. 100% |  52.7 MiB/s |   1.4 MiB |  00m00s
[ 57/187] Installing libmount-0:2.41.3- 100% | 190.3 MiB/s | 389.7 KiB |  00m00s
[ 58/187] Installing gnupg2-verify-0:2. 100% |  18.6 MiB/s | 361.7 KiB |  00m00s
[ 59/187] Installing dwz-0:0.16-3.fc44. 100% |  15.9 MiB/s | 292.3 KiB |  00m00s
[ 60/187] Installing mpfr-0:4.2.2-3.fc4 100% | 276.9 MiB/s | 850.8 KiB |  00m00s
[ 61/187] Installing gawk-0:5.4.0-2.fc4 100% |  89.0 MiB/s |   2.2 MiB |  00m00s
[ 62/187] Installing libksba-0:1.6.8-1. 100% | 207.2 MiB/s | 424.4 KiB |  00m00s
[ 63/187] Installing unzip-0:6.0-69.fc4 100% |  23.1 MiB/s | 449.3 KiB |  00m00s
[ 64/187] Installing file-libs-0:5.47-1 100% | 609.5 MiB/s |  12.2 MiB |  00m00s
[ 65/187] Installing file-0:5.47-1.fc45 100% |   5.3 MiB/s | 102.6 KiB |  00m00s
[ 66/187] Installing diffutils-0:3.12-5 100% |  68.4 MiB/s |   1.6 MiB |  00m00s
[ 67/187] Installing libeconf-0:0.7.9-3 100% |  64.9 MiB/s |  66.4 KiB |  00m00s
[ 68/187] Installing libcap-ng-0:0.9.1- 100% |  68.9 MiB/s |  70.6 KiB |  00m00s
[ 69/187] Installing audit-libs-0:4.1.3 100% | 191.9 MiB/s | 393.1 KiB |  00m00s
[ 70/187] Installing pam-libs-0:1.7.2-1 100% | 129.8 MiB/s | 132.9 KiB |  00m00s
[ 71/187] Installing libcap-0:2.77-2.fc 100% |  11.2 MiB/s | 217.2 KiB |  00m00s
[ 72/187] Installing libsemanage-0:3.10 100% | 153.4 MiB/s | 314.1 KiB |  00m00s
[ 73/187] Installing libsmartcols-0:2.4 100% | 184.9 MiB/s | 189.3 KiB |  00m00s
[ 74/187] Installing lua-libs-0:5.5.0-1 100% | 292.4 MiB/s | 299.4 KiB |  00m00s
[ 75/187] Installing json-c-0:0.18-8.fc 100% |  81.9 MiB/s |  83.9 KiB |  00m00s
[ 76/187] Installing libffi-0:3.5.2-2.f 100% |  87.0 MiB/s |  89.1 KiB |  00m00s
[ 77/187] Installing p11-kit-0:0.26.2-1 100% |  96.1 MiB/s |   2.6 MiB |  00m00s
[ 78/187] Installing alternatives-0:1.3 100% |   3.5 MiB/s |  63.6 KiB |  00m00s
[ 79/187] Installing p11-kit-trust-0:0. 100% |  18.0 MiB/s | 480.0 KiB |  00m00s
[ 80/187] Installing ngtcp2-0:1.19.0-2. 100% | 162.0 MiB/s | 331.7 KiB |  00m00s
[ 81/187] Installing openssl-libs-1:3.5 100% | 367.5 MiB/s |   9.2 MiB |  00m00s
[ 82/187] Installing coreutils-0:9.10-3 100% | 138.8 MiB/s |   5.7 MiB |  00m00s
[ 83/187] Installing ca-certificates-0: 100% |   2.0 MiB/s |   2.5 MiB |  00m01s
[ 84/187] Installing gzip-0:1.14-2.fc44 100% |  18.9 MiB/s | 407.1 KiB |  00m00s
[ 85/187] Installing rpm-sequoia-0:1.10 100% | 337.8 MiB/s |   2.4 MiB |  00m00s
[ 86/187] Installing libfsverity-0:1.7- 100% |  28.7 MiB/s |  29.4 KiB |  00m00s
[ 87/187] Installing libevent-0:2.1.12- 100% | 319.8 MiB/s | 982.4 KiB |  00m00s
[ 88/187] Installing systemd-standalone 100% |  41.1 MiB/s | 841.8 KiB |  00m00s
[ 89/187] Installing rpm-libs-0:6.0.1-5 100% | 313.4 MiB/s | 962.8 KiB |  00m00s
[ 90/187] Installing ngtcp2-crypto-ossl 100% |  51.2 MiB/s |  52.5 KiB |  00m00s
[ 91/187] Installing util-linux-core-0: 100% |  61.8 MiB/s |   1.5 MiB |  00m00s
[ 92/187] Installing zip-0:3.0-45.fc44. 100% |  34.3 MiB/s | 701.9 KiB |  00m00s
[ 93/187] Installing gnupg2-keyboxd-0:2 100% |  28.2 MiB/s | 202.5 KiB |  00m00s
[ 94/187] Installing libpsl-0:0.21.5-7. 100% |  75.6 MiB/s |  77.4 KiB |  00m00s
[ 95/187] Installing tar-2:1.35-8.fc44. 100% | 102.7 MiB/s |   3.0 MiB |  00m00s
[ 96/187] Installing linkdupes-0:0.7.3- 100% |  38.2 MiB/s | 782.3 KiB |  00m00s
[ 97/187] Installing libselinux-utils-0 100% |  15.6 MiB/s | 320.1 KiB |  00m00s
[ 98/187] Installing liblastlog2-0:2.41 100% |   6.5 MiB/s |  39.8 KiB |  00m00s
[ 99/187] Installing libfdisk-0:2.41.3- 100% | 126.8 MiB/s | 389.4 KiB |  00m00s
[100/187] Installing util-linux-0:2.41. 100% |  79.6 MiB/s |   3.6 MiB |  00m00s
[101/187] Installing policycoreutils-0: 100% |  27.9 MiB/s | 912.8 KiB |  00m00s
[102/187] Installing selinux-policy-0:4 100% |   1.6 MiB/s |  33.6 KiB |  00m00s
[103/187] Installing selinux-policy-tar 100% | 178.1 MiB/s |  14.8 MiB |  00m00s
[104/187] Installing zstd-0:1.5.7-5.fc4 100% |  22.5 MiB/s | 506.0 KiB |  00m00s
[105/187] Installing libusb1-0:1.0.29-5 100% |  19.2 MiB/s | 176.9 KiB |  00m00s
>>> Running sysusers scriptlet: tpm2-tss-0:4.1.3-9.fc44.x86_64
>>> Finished sysusers scriptlet: tpm2-tss-0:4.1.3-9.fc44.x86_64
>>> Scriptlet output:
>>> Creating group 'tss' with GID 59.
>>> Creating user 'tss' (Account used for TPM access) with UID 59 and GID 59.
>>> 
[106/187] Installing tpm2-tss-0:4.1.3-9 100% | 233.3 MiB/s |   1.6 MiB |  00m00s
[107/187] Installing ima-evm-utils-libs 100% |  60.5 MiB/s |  61.9 KiB |  00m00s
[108/187] Installing gnupg2-gpg-agent-0 100% |  23.9 MiB/s | 686.5 KiB |  00m00s
[109/187] Installing libxml2-0:2.12.10- 100% |  73.8 MiB/s |   1.8 MiB |  00m00s
[110/187] Installing nettle-0:3.10.1-3. 100% | 259.6 MiB/s | 797.4 KiB |  00m00s
[111/187] Installing gnutls-0:3.8.12-1. 100% | 339.6 MiB/s |   3.7 MiB |  00m00s
[112/187] Installing bzip2-0:1.0.8-23.f 100% |   5.1 MiB/s |  99.5 KiB |  00m00s
[113/187] Installing add-determinism-0: 100% |  94.9 MiB/s |   2.2 MiB |  00m00s
[114/187] Installing build-reproducibil 100% |   1.5 MiB/s |   1.5 KiB |  00m00s
[115/187] Installing cpio-0:2.15-9.fc44 100% |  52.5 MiB/s |   1.1 MiB |  00m00s
[116/187] Installing ed-0:1.22.5-2.fc45 100% |   7.8 MiB/s | 152.0 KiB |  00m00s
[117/187] Installing patch-0:2.8-4.fc44 100% |  11.7 MiB/s | 228.1 KiB |  00m00s
[118/187] Installing lz4-libs-0:1.10.0- 100% | 154.6 MiB/s | 158.4 KiB |  00m00s
[119/187] Installing libarchive-0:3.8.6 100% | 324.6 MiB/s | 997.2 KiB |  00m00s
[120/187] Installing libtool-ltdl-0:2.5 100% |  69.5 MiB/s |  71.1 KiB |  00m00s
[121/187] Installing gdbm-libs-1:1.23-1 100% | 128.3 MiB/s | 131.3 KiB |  00m00s
[122/187] Installing cyrus-sasl-lib-0:2 100% |  96.3 MiB/s |   2.3 MiB |  00m00s
[123/187] Installing openldap-0:2.6.13- 100% | 219.3 MiB/s | 673.7 KiB |  00m00s
[124/187] Installing gnupg2-dirmngr-0:2 100% |  23.0 MiB/s | 636.6 KiB |  00m00s
[125/187] Installing gnupg2-0:2.4.9-5.f 100% | 171.3 MiB/s |   6.5 MiB |  00m00s
[126/187] Installing rpm-sign-libs-0:6. 100% |  39.4 MiB/s |  40.3 KiB |  00m00s
[127/187] Installing gpgverify-0:2.2-4. 100% |   0.0   B/s |   9.4 KiB |  00m00s
[128/187] Installing libgomp-0:16.0.1-0 100% | 280.3 MiB/s | 574.1 KiB |  00m00s
[129/187] Installing jansson-0:2.14-4.f 100% |  88.2 MiB/s |  90.3 KiB |  00m00s
[130/187] Installing libpkgconf-0:2.5.1 100% |  89.1 MiB/s |  91.3 KiB |  00m00s
[131/187] Installing pkgconf-0:2.5.1-1. 100% |   4.9 MiB/s |  95.2 KiB |  00m00s
[132/187] Installing pkgconf-pkg-config 100% |  98.5 KiB/s |   1.8 KiB |  00m00s
[133/187] Installing xxhash-libs-0:0.8. 100% |  93.2 MiB/s |  95.4 KiB |  00m00s
[134/187] Installing libbrotli-0:1.2.0- 100% | 282.3 MiB/s | 867.3 KiB |  00m00s
[135/187] Installing libnghttp2-0:1.68. 100% | 163.3 MiB/s | 167.3 KiB |  00m00s
[136/187] Installing libnghttp3-0:1.15. 100% | 156.8 MiB/s | 160.6 KiB |  00m00s
[137/187] Installing keyutils-libs-0:1. 100% |  54.3 MiB/s |  55.6 KiB |  00m00s
[138/187] Installing libcom_err-0:1.47. 100% |  62.5 MiB/s |  64.0 KiB |  00m00s
[139/187] Installing libverto-0:0.3.2-1 100% |  26.4 MiB/s |  27.1 KiB |  00m00s
[140/187] Installing krb5-libs-0:1.22.2 100% | 299.8 MiB/s |   2.4 MiB |  00m00s
[141/187] Installing libcbor-0:0.13.0-2 100% |  79.0 MiB/s |  80.9 KiB |  00m00s
[142/187] Installing libfido2-0:1.16.0- 100% | 242.1 MiB/s | 247.9 KiB |  00m00s
[143/187] Installing libssh-0:0.12.0-1. 100% | 234.8 MiB/s | 721.2 KiB |  00m00s
[144/187] Installing libcurl-0:8.19.0~r 100% | 332.7 MiB/s |   1.0 MiB |  00m00s
[145/187] Installing curl-0:8.19.0~rc3- 100% |  17.0 MiB/s | 486.0 KiB |  00m00s
[146/187] Installing rpm-0:6.0.1-5.fc45 100% |  60.7 MiB/s |   2.6 MiB |  00m00s
[147/187] Installing cmake-srpm-macros- 100% |   0.0   B/s | 804.0   B |  00m00s
[148/187] Installing efi-srpm-macros-0: 100% |  40.2 MiB/s |  41.2 KiB |  00m00s
[149/187] Installing java-srpm-macros-0 100% |   0.0   B/s |   1.1 KiB |  00m00s
[150/187] Installing lua-srpm-macros-0: 100% |   0.0   B/s |   1.9 KiB |  00m00s
[151/187] Installing tree-sitter-srpm-m 100% |   0.0   B/s |   9.3 KiB |  00m00s
[152/187] Installing zig-srpm-macros-0: 100% |   0.0   B/s |   1.9 KiB |  00m00s
[153/187] Installing filesystem-srpm-ma 100% |   0.0   B/s |  38.9 KiB |  00m00s
[154/187] Installing elfutils-default-y 100% | 408.6 KiB/s |   2.0 KiB |  00m00s
[155/187] Installing elfutils-libs-0:0. 100% | 233.4 MiB/s | 717.1 KiB |  00m00s
[156/187] Installing elfutils-debuginfo 100% |   4.0 MiB/s |  86.0 KiB |  00m00s
[157/187] Installing elfutils-0:0.194-5 100% | 109.6 MiB/s |   3.0 MiB |  00m00s
[158/187] Installing binutils-0:2.46.50 100% | 294.7 MiB/s |  28.3 MiB |  00m00s
[159/187] Installing gdb-minimal-0:17.1 100% | 244.7 MiB/s |  14.2 MiB |  00m00s
[160/187] Installing debugedit-0:5.3-2. 100% |  10.9 MiB/s | 224.1 KiB |  00m00s
[161/187] Installing rpm-build-libs-0:6 100% | 135.3 MiB/s | 277.2 KiB |  00m00s
[162/187] Installing rust-srpm-macros-0 100% |   0.0   B/s |   6.4 KiB |  00m00s
[163/187] Installing qt6-srpm-macros-0: 100% | 146.1 KiB/s | 748.0   B |  00m00s
[164/187] Installing qt5-srpm-macros-0: 100% |   0.0   B/s | 776.0   B |  00m00s
[165/187] Installing perl-srpm-macros-0 100% |   0.0   B/s |   1.1 KiB |  00m00s
[166/187] Installing package-notes-srpm 100% |   0.0   B/s |   2.1 KiB |  00m00s
[167/187] Installing openblas-srpm-macr 100% |   0.0   B/s | 392.0   B |  00m00s
[168/187] Installing ocaml-srpm-macros- 100% |   0.0   B/s |   2.1 KiB |  00m00s
[169/187] Installing kernel-srpm-macros 100% |   0.0   B/s |   2.3 KiB |  00m00s
[170/187] Installing gnat-srpm-macros-0 100% |   0.0   B/s |   1.3 KiB |  00m00s
[171/187] Installing ghc-srpm-macros-0: 100% |   0.0   B/s |   1.0 KiB |  00m00s
[172/187] Installing gap-srpm-macros-0: 100% |   0.0   B/s |   2.7 KiB |  00m00s
[173/187] Installing fpc-srpm-macros-0: 100% |   0.0   B/s | 420.0   B |  00m00s
[174/187] Installing ansible-srpm-macro 100% |   0.0   B/s |  36.2 KiB |  00m00s
[175/187] Installing redhat-rpm-config- 100% |  92.7 MiB/s | 189.9 KiB |  00m00s
[176/187] Installing forge-srpm-macros- 100% |  39.3 MiB/s |  40.3 KiB |  00m00s
[177/187] Installing rpm-build-0:6.0.1- 100% |  14.1 MiB/s | 303.7 KiB |  00m00s
[178/187] Installing erlang-srpm-macros 100% |   0.0   B/s |   2.5 KiB |  00m00s
[179/187] Installing pyproject-srpm-mac 100% |   0.0   B/s |   2.5 KiB |  00m00s
[180/187] Installing fonts-srpm-macros- 100% |  55.7 MiB/s |  57.0 KiB |  00m00s
[181/187] Installing go-srpm-macros-0:3 100% |  61.6 MiB/s |  63.0 KiB |  00m00s
[182/187] Installing R-srpm-macros-0:1. 100% |   0.0   B/s |   4.4 KiB |  00m00s
[183/187] Installing python-srpm-macros 100% |  51.7 MiB/s |  52.9 KiB |  00m00s
[184/187] Installing rpm-plugin-selinux 100% |   0.0   B/s |  12.9 KiB |  00m00s
[185/187] Installing which-0:2.23-4.fc4 100% |   4.2 MiB/s |  85.6 KiB |  00m00s
[186/187] Installing shadow-utils-2:4.1 100% | 111.7 MiB/s |   4.0 MiB |  00m00s
[187/187] Installing info-0:7.3-1.fc45. 100% |  49.9 KiB/s | 372.8 KiB |  00m07s
Warning: skipped OpenPGP checks for 4 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: R-srpm-macros-1.3.5-1.fc45.noarch
add-determinism-0.7.3-1.fc45.x86_64
alternatives-1.33-5.fc44.x86_64
ansible-srpm-macros-1-20.1.fc44.noarch
audit-libs-4.1.3-1.fc44.x86_64
bash-5.3.9-3.fc44.x86_64
binutils-2.46.50-3.fc45.x86_64
build-reproducibility-srpm-macros-0.7.3-1.fc45.noarch
bzip2-1.0.8-23.fc44.x86_64
bzip2-libs-1.0.8-23.fc44.x86_64
ca-certificates-2025.2.80_v9.0.304-6.fc45.noarch
cmake-srpm-macros-4.2.3-2.fc45.noarch
coreutils-9.10-3.fc45.x86_64
coreutils-common-9.10-3.fc45.x86_64
cpio-2.15-9.fc44.x86_64
crypto-policies-20251128-3.git19878fe.fc44.noarch
curl-8.19.0~rc3-1.fc45.x86_64
cyrus-sasl-lib-2.1.28-35.fc44.x86_64
debugedit-5.3-2.fc45.x86_64
diffutils-3.12-5.fc44.x86_64
dwz-0.16-3.fc44.x86_64
ed-1.22.5-2.fc45.x86_64
efi-srpm-macros-6-6.fc44.noarch
elfutils-0.194-5.fc45.x86_64
elfutils-debuginfod-client-0.194-5.fc45.x86_64
elfutils-default-yama-scope-0.194-5.fc45.noarch
elfutils-libelf-0.194-5.fc45.x86_64
elfutils-libs-0.194-5.fc45.x86_64
erlang-srpm-macros-0.3.11-1.fc45.noarch
fedora-gpg-keys-45-0.1.noarch
fedora-release-45-0.2.noarch
fedora-release-common-45-0.2.noarch
fedora-release-identity-basic-45-0.2.noarch
fedora-repos-45-0.1.noarch
fedora-repos-rawhide-45-0.1.noarch
file-5.47-1.fc45.x86_64
file-libs-5.47-1.fc45.x86_64
filesystem-3.18-56.fc45.x86_64
filesystem-srpm-macros-3.18-56.fc45.noarch
findutils-4.10.0-7.fc44.x86_64
fonts-srpm-macros-5.0.0-3.fc45.noarch
forge-srpm-macros-0.4.0-4.fc44.noarch
fpc-srpm-macros-1.3-16.fc44.noarch
gap-srpm-macros-2-2.fc44.noarch
gawk-5.4.0-2.fc45.x86_64
gdb-minimal-17.1-5.fc45.x86_64
gdbm-libs-1.23-11.fc44.x86_64
ghc-srpm-macros-1.10-1.fc44.noarch
glibc-2.43.9000-4.fc45.x86_64
glibc-common-2.43.9000-4.fc45.x86_64
glibc-gconv-extra-2.43.9000-4.fc45.x86_64
glibc-minimal-langpack-2.43.9000-4.fc45.x86_64
gmp-6.3.0-5.fc44.x86_64
gnat-srpm-macros-7-2.fc44.noarch
gnulib-l10n-20241231-2.fc44.noarch
gnupg2-2.4.9-5.fc44.x86_64
gnupg2-dirmngr-2.4.9-5.fc44.x86_64
gnupg2-gpg-agent-2.4.9-5.fc44.x86_64
gnupg2-gpgconf-2.4.9-5.fc44.x86_64
gnupg2-keyboxd-2.4.9-5.fc44.x86_64
gnupg2-verify-2.4.9-5.fc44.x86_64
gnutls-3.8.12-1.fc45.x86_64
go-srpm-macros-3.8.0-2.fc44.noarch
gpg-pubkey-36f612dcf27f7d1a48a835e4dbfcf71c6d9f90a6-6786af3b
gpg-pubkey-4f50a6114cd5c6976a7f1179655a4b02f577861e-6888bc98
gpg-pubkey-d924b10d3e810dabdd8b56b596e7e91491211fce-697c9899
gpgverify-2.2-4.fc44.noarch
grep-3.12-3.fc44.x86_64
gzip-1.14-2.fc44.x86_64
ima-evm-utils-libs-1.6.2-10.fc45.x86_64
info-7.3-1.fc45.x86_64
jansson-2.14-4.fc44.x86_64
java-srpm-macros-1-8.fc44.noarch
json-c-0.18-8.fc44.x86_64
kernel-srpm-macros-1.0-28.fc44.noarch
keyutils-libs-1.6.3-7.fc44.x86_64
krb5-libs-1.22.2-2.fc45.x86_64
libacl-2.3.2-6.fc44.x86_64
libarchive-3.8.6-1.fc45.x86_64
libassuan-2.5.7-5.fc44.x86_64
libattr-2.5.2-8.fc44.x86_64
libblkid-2.41.3-12.fc44.x86_64
libbrotli-1.2.0-3.fc44.x86_64
libcap-2.77-2.fc44.x86_64
libcap-ng-0.9.1-1.fc45.x86_64
libcbor-0.13.0-2.fc44.x86_64
libcom_err-1.47.4-1.fc45.x86_64
libcurl-8.19.0~rc3-1.fc45.x86_64
libeconf-0.7.9-3.fc44.x86_64
libevent-2.1.12-17.fc44.x86_64
libfdisk-2.41.3-12.fc44.x86_64
libffi-3.5.2-2.fc44.x86_64
libfido2-1.16.0-5.fc44.x86_64
libfsverity-1.7-1.fc45.x86_64
libgcc-16.0.1-0.9.fc45.x86_64
libgcrypt-1.12.1-1.fc45.x86_64
libgomp-16.0.1-0.9.fc45.x86_64
libgpg-error-1.59-1.fc45.x86_64
libidn2-2.3.8-3.fc44.x86_64
libksba-1.6.8-1.fc45.x86_64
liblastlog2-2.41.3-12.fc44.x86_64
libmount-2.41.3-12.fc44.x86_64
libnghttp2-1.68.0-3.fc44.x86_64
libnghttp3-1.15.0-1.fc44.x86_64
libpkgconf-2.5.1-1.fc45.x86_64
libpsl-0.21.5-7.fc44.x86_64
libselinux-3.10-1.fc44.x86_64
libselinux-utils-3.10-1.fc44.x86_64
libsemanage-3.10-1.fc44.x86_64
libsepol-3.10-1.fc44.x86_64
libsmartcols-2.41.3-12.fc44.x86_64
libssh-0.12.0-1.fc45.x86_64
libssh-config-0.12.0-1.fc45.noarch
libstdc++-16.0.1-0.9.fc45.x86_64
libtasn1-4.21.0-1.fc45.x86_64
libtool-ltdl-2.5.4-10.fc44.x86_64
libunistring-1.1-11.fc44.x86_64
libusb1-1.0.29-5.fc44.x86_64
libuuid-2.41.3-12.fc44.x86_64
libverto-0.3.2-12.fc44.x86_64
libxcrypt-4.5.2-3.fc44.x86_64
libxml2-2.12.10-6.fc44.x86_64
libzstd-1.5.7-5.fc44.x86_64
linkdupes-0.7.3-1.fc45.x86_64
lua-libs-5.5.0-1.fc45.x86_64
lua-srpm-macros-1-17.fc44.noarch
lz4-libs-1.10.0-4.fc44.x86_64
mpfr-4.2.2-3.fc44.x86_64
ncurses-base-6.6-1.fc44.noarch
ncurses-libs-6.6-1.fc44.x86_64
nettle-3.10.1-3.fc44.x86_64
ngtcp2-1.19.0-2.fc44.x86_64
ngtcp2-crypto-ossl-1.19.0-2.fc44.x86_64
npth-1.8-4.fc44.x86_64
ocaml-srpm-macros-11-3.fc44.noarch
openblas-srpm-macros-2-21.fc44.noarch
openldap-2.6.13-1.fc45.x86_64
openssl-libs-3.5.5-1.fc44.x86_64
p11-kit-0.26.2-1.fc45.x86_64
p11-kit-trust-0.26.2-1.fc45.x86_64
package-notes-srpm-macros-0.17-3.fc45.noarch
pam-libs-1.7.2-1.fc44.x86_64
patch-2.8-4.fc44.x86_64
pcre2-10.47-1.fc44.1.x86_64
pcre2-syntax-10.47-1.fc44.1.noarch
perl-srpm-macros-1-61.fc44.noarch
pkgconf-2.5.1-1.fc45.x86_64
pkgconf-m4-2.5.1-1.fc45.noarch
pkgconf-pkg-config-2.5.1-1.fc45.x86_64
policycoreutils-3.10-2.fc45.x86_64
popt-1.19-10.fc44.x86_64
publicsuffix-list-dafsa-20260116-1.fc44.noarch
pyproject-srpm-macros-1.18.7-1.fc45.noarch
python-srpm-macros-3.14-11.fc45.noarch
qt5-srpm-macros-5.15.18-2.fc44.noarch
qt6-srpm-macros-6.10.2-1.fc45.noarch
readline-8.3-4.fc44.x86_64
redhat-rpm-config-344-1.fc45.noarch
rpm-6.0.1-5.fc45.x86_64
rpm-build-6.0.1-5.fc45.x86_64
rpm-build-libs-6.0.1-5.fc45.x86_64
rpm-libs-6.0.1-5.fc45.x86_64
rpm-plugin-selinux-6.0.1-5.fc45.x86_64
rpm-sequoia-1.10.1-1.fc45.x86_64
rpm-sign-libs-6.0.1-5.fc45.x86_64
rust-srpm-macros-28.4-3.fc44.noarch
sed-4.9-8.fc45.x86_64
selinux-policy-43.1-1.fc45.noarch
selinux-policy-targeted-43.1-1.fc45.noarch
setup-2.15.0-29.fc45.noarch
shadow-utils-4.19.3-1.fc45.x86_64
sqlite-libs-3.52.0-1.fc45.x86_64
systemd-libs-260~rc4-1.fc45.x86_64
systemd-standalone-sysusers-260~rc4-1.fc45.x86_64
tar-1.35-8.fc44.x86_64
tpm2-tss-4.1.3-9.fc44.x86_64
tree-sitter-srpm-macros-0.4.2-2.fc44.noarch
unzip-6.0-69.fc44.x86_64
util-linux-2.41.3-12.fc44.x86_64
util-linux-core-2.41.3-12.fc44.x86_64
which-2.23-4.fc44.x86_64
xxhash-libs-0.8.3-4.fc44.x86_64
xz-5.8.2-2.fc44.x86_64
xz-libs-5.8.2-2.fc44.x86_64
zig-srpm-macros-1-8.fc44.noarch
zip-3.0-45.fc44.x86_64
zlib-ng-compat-2.3.3-5.fc45.x86_64
zstd-1.5.7-5.fc44.x86_64
Start: buildsrpm
Start: rpmbuild -bs
Building target platforms: x86_64
Building for target x86_64
setting SOURCE_DATE_EPOCH=1773792000
Wrote: /builddir/build/SRPMS/llama-cpp-b8064-1.fc45.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_64-1773846985.598827/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-uie1_oaj/llama-cpp/llama-cpp.spec) Config(child) 0 minutes 28 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-b8064-1.fc45.src.rpm)  Config(fedora-rawhide-x86_64)
Start(bootstrap): chroot init
INFO: mounting tmpfs at /var/lib/mock/fedora-rawhide-x86_64-bootstrap-1773846985.598827/root.
INFO: reusing tmpfs at /var/lib/mock/fedora-rawhide-x86_64-bootstrap-1773846985.598827/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_64-1773846985.598827/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-6.0.1-5.fc45.x86_64
  rpm-sequoia-1.10.1-1.fc45.x86_64
  dnf5-5.4.0.0-5.fc45.x86_64
  dnf5-plugins-5.4.0.0-5.fc45.x86_64
Finish: chroot init
Start: build phase for llama-cpp-b8064-1.fc45.src.rpm
Start: build setup for llama-cpp-b8064-1.fc45.src.rpm
Building target platforms: x86_64
Building for target x86_64
setting SOURCE_DATE_EPOCH=1773792000
Wrote: /builddir/build/SRPMS/llama-cpp-b8064-1.fc45.src.rpm
Updating and loading repositories:
 fedora                                 100% |  66.8 KiB/s |  21.6 KiB |  00m00s
 Copr repository                        100% |  95.7 KiB/s |   1.5 KiB |  00m00s
Repositories loaded.
Package "curl-8.19.0~rc3-1.fc45.x86_64" is already installed.

Package                         Arch   Version                          Repository      Size
Installing:
 cmake                          x86_64 0:4.2.3-2.fc45                   fedora      39.0 MiB
 gcc-c++                        x86_64 0:16.0.1-0.9.fc45                copr_base   46.3 MiB
 git                            x86_64 0:2.53.0-1.fc44                  fedora      56.4 KiB
 hipblas-devel                  x86_64 0:7.2.0-4.fc45                   copr_base    2.4 MiB
 langpacks-en                   noarch 0:4.3-1.fc45                     fedora     400.0   B
 libcurl-devel                  x86_64 0:8.19.0~rc3-1.fc45              fedora       1.4 MiB
 openmpi                        x86_64 0:5.0.10-1.fc45                  fedora       7.1 MiB
 pthreadpool-devel              x86_64 0:0.0^git20230829.4fe0e1e-9.fc44 fedora      99.1 KiB
 rocblas-devel                  x86_64 0:7.2.0-5.fc45                   copr_base    2.7 MiB
 rocm-comgr-devel               x86_64 0:22-5.rocm7.2.0.fc45            copr_base   99.2 KiB
 rocm-hip-devel                 x86_64 0:7.2.0-3.fc45                   copr_base    2.4 MiB
 rocm-rpm-macros                noarch 0:7.2.0-1.fc45                   fedora      18.9 KiB
 rocm-runtime-devel             x86_64 0:7.2.0-3.fc45                   copr_base  683.9 KiB
 wget2-wget                     x86_64 0:2.2.1-2.fc44                   fedora      42.0   B
 xxd                            x86_64 2:9.2.148-1.fc45                 fedora      33.1 KiB
Installing dependencies:
 abattis-cantarell-vf-fonts     noarch 0:0.301-17.fc44                  fedora     192.7 KiB
 annobin-docs                   noarch 0:13.07-1.fc45                   fedora      99.2 KiB
 annobin-plugin-gcc             x86_64 0:13.07-1.fc45                   fedora     697.1 KiB
 brotli                         x86_64 0:1.2.0-3.fc44                   fedora      33.4 KiB
 brotli-devel                   x86_64 0:1.2.0-3.fc44                   fedora      65.9 KiB
 cmake-data                     noarch 0:4.2.3-2.fc45                   fedora       9.3 MiB
 cmake-filesystem               x86_64 0:4.2.3-2.fc45                   fedora       0.0   B
 cmake-rpm-macros               noarch 0:4.2.3-2.fc45                   fedora       8.4 KiB
 cpp                            x86_64 0:16.0.1-0.9.fc45                copr_base   42.1 MiB
 default-fonts-core-sans        noarch 0:4.3-1.fc45                     fedora      11.9 KiB
 dns-root-data                  noarch 0:2026260100-2.fc44              fedora      10.9 KiB
 emacs-filesystem               x86_64 1:30.2-2.fc44                    fedora       0.0   B
 environment-modules            x86_64 0:5.6.1-3.fc44                   fedora       1.9 MiB
 expat                          x86_64 0:2.7.4-1.fc45                   fedora     321.2 KiB
 fonts-filesystem               noarch 1:5.0.0-3.fc45                   fedora       0.0   B
 gcc                            x86_64 0:16.0.1-0.9.fc45                copr_base  120.4 MiB
 gcc-plugin-annobin             x86_64 0:16.0.1-0.9.fc45                copr_base   57.2 KiB
 git-core                       x86_64 0:2.53.0-1.fc44                  fedora      24.6 MiB
 git-core-doc                   noarch 0:2.53.0-1.fc44                  fedora      18.6 MiB
 glibc-devel                    x86_64 0:2.43.9000-4.fc45               fedora       2.3 MiB
 gnutls-dane                    x86_64 0:3.8.12-1.fc45                  fedora      60.8 KiB
 google-noto-fonts-common       noarch 0:20260201-2.fc45                fedora      17.7 KiB
 google-noto-sans-mono-vf-fonts noarch 0:20260201-2.fc45                fedora     561.2 KiB
 google-noto-sans-vf-fonts      noarch 0:20260201-2.fc45                fedora       1.4 MiB
 google-noto-serif-vf-fonts     noarch 0:20260201-2.fc45                fedora       1.6 MiB
 gpgme                          x86_64 0:2.0.1-3.fc44                   fedora     617.8 KiB
 groff-base                     x86_64 0:1.24.0-1.fc45                  fedora       4.2 MiB
 hipblas                        x86_64 0:7.2.0-4.fc45                   copr_base  803.0 KiB
 hipblas-common-devel           noarch 0:7.2.0-2.fc45                   copr_base   16.8 KiB
 hipcc                          x86_64 0:22-5.rocm7.2.0.fc45            copr_base  632.8 KiB
 hiredis                        x86_64 0:1.2.0-8.fc44                   fedora     105.7 KiB
 hwdata                         noarch 0:0.405-1.fc45                   fedora       9.8 MiB
 hwloc-libs                     x86_64 0:2.12.0-3.fc44                  fedora       2.9 MiB
 jsoncpp                        x86_64 0:1.9.6-3.fc44                   fedora     285.6 KiB
 kernel-headers                 x86_64 0:7.0.0-0.rc4.36.fc45            fedora       6.9 MiB
 keyutils-libs-devel            x86_64 0:1.6.3-7.fc44                   fedora      48.2 KiB
 kmod                           x86_64 0:34.2-4.fc44                    fedora     251.0 KiB
 krb5-devel                     x86_64 0:1.22.2-2.fc45                  fedora     709.9 KiB
 langpacks-core-en              noarch 0:4.3-1.fc45                     fedora     398.0   B
 langpacks-fonts-en             noarch 0:4.3-1.fc45                     fedora     341.0   B
 less                           x86_64 0:692-3.fc45                     fedora     465.3 KiB
 libatomic                      x86_64 0:16.0.1-0.9.fc45                copr_base   45.5 KiB
 libcom_err-devel               x86_64 0:1.47.4-1.fc45                  fedora      16.7 KiB
 libdrm                         x86_64 0:2.4.131-1.fc44                 fedora     404.1 KiB
 libedit                        x86_64 0:3.1-58.20251016cvs.fc44        fedora     248.1 KiB
 libfabric                      x86_64 0:2.3.1-3.fc45                   fedora       9.3 MiB
 libgfortran                    x86_64 0:16.0.1-0.9.fc45                copr_base    3.4 MiB
 libibverbs                     x86_64 0:61.0-2.fc44                    fedora       1.3 MiB
 libidn2-devel                  x86_64 0:2.3.8-3.fc44                   fedora     149.1 KiB
 libkadm5                       x86_64 0:1.22.2-2.fc45                  fedora     217.8 KiB
 libmpc                         x86_64 0:1.3.1-9.fc44                   fedora     160.4 KiB
 libnghttp2-devel               x86_64 0:1.68.0-3.fc44                  fedora     288.0 KiB
 libnghttp3-devel               x86_64 0:1.15.0-1.fc44                  fedora     107.4 KiB
 libnl3                         x86_64 0:3.12.0-3.fc44                  fedora       1.1 MiB
 libpciaccess                   x86_64 0:0.16-17.fc44                   fedora      48.4 KiB
 libpipeline                    x86_64 0:1.5.8-4.fc44                   fedora     145.0 KiB
 libpsl-devel                   x86_64 0:0.21.5-7.fc44                  fedora     110.2 KiB
 libpsm2                        x86_64 0:12.0.1-7.fc45                  fedora     454.2 KiB
 libquadmath                    x86_64 0:16.0.1-0.9.fc45                copr_base  325.9 KiB
 librdmacm                      x86_64 0:61.0-2.fc44                    fedora     153.8 KiB
 libselinux-devel               x86_64 0:3.10-1.fc44                    fedora     122.2 KiB
 libsepol-devel                 x86_64 0:3.10-1.fc44                    fedora     121.5 KiB
 libssh-devel                   x86_64 0:0.12.0-1.fc45                  fedora     203.5 KiB
 libstdc++-devel                x86_64 0:16.0.1-0.9.fc45                copr_base   38.8 MiB
 libtommath                     x86_64 0:1.3.1~rc1-7.fc44               fedora     130.2 KiB
 libuv                          x86_64 1:1.52.1-1.fc45                  fedora     593.2 KiB
 libverto-devel                 x86_64 0:0.3.2-12.fc44                  fedora      25.7 KiB
 libxcrypt-devel                x86_64 0:4.5.2-3.fc44                   fedora      31.0 KiB
 make                           x86_64 1:4.4.1-12.fc44                  fedora       1.8 MiB
 man-db                         x86_64 0:2.13.1-3.fc44                  fedora       2.9 MiB
 mpdecimal                      x86_64 0:4.0.1-3.fc44                   fedora     217.1 KiB
 munge-libs                     x86_64 0:0.5.18-1.fc45                  fedora      31.9 KiB
 ncurses                        x86_64 0:6.6-1.fc44                     fedora     611.4 KiB
 ngtcp2-crypto-ossl-devel       x86_64 0:1.19.0-2.fc44                  fedora       7.5 KiB
 ngtcp2-devel                   x86_64 0:1.19.0-2.fc44                  fedora     282.4 KiB
 ninja-build                    x86_64 0:1.13.2-2.fc44                  fedora     496.5 KiB
 numactl-libs                   x86_64 0:2.0.19-4.fc44                  fedora      56.8 KiB
 openssh                        x86_64 0:10.2p1-5.fc45                  fedora       1.4 MiB
 openssh-clients                x86_64 0:10.2p1-5.fc45                  fedora       2.7 MiB
 openssl-devel                  x86_64 1:3.5.5-1.fc44                   fedora       4.5 MiB
 orangefs                       x86_64 0:2.10.1-3.fc45                  fedora       1.7 MiB
 pcre2-devel                    x86_64 0:10.47-1.fc44.1                 fedora       2.1 MiB
 pcre2-utf16                    x86_64 0:10.47-1.fc44.1                 fedora     655.2 KiB
 pcre2-utf32                    x86_64 0:10.47-1.fc44.1                 fedora     623.1 KiB
 perl-AutoLoader                noarch 0:5.74-523.fc45                  fedora      20.6 KiB
 perl-B                         x86_64 0:1.89-523.fc45                  fedora     501.2 KiB
 perl-Carp                      noarch 0:1.54-521.fc44                  fedora      46.6 KiB
 perl-Class-Struct              noarch 0:0.68-523.fc45                  fedora      25.4 KiB
 perl-Data-Dumper               x86_64 0:2.191-522.fc44                 fedora     115.5 KiB
 perl-Digest                    noarch 0:1.20-521.fc44                  fedora      35.3 KiB
 perl-Digest-MD5                x86_64 0:2.59-521.fc44                  fedora      59.6 KiB
 perl-DynaLoader                x86_64 0:1.57-523.fc45                  fedora      32.1 KiB
 perl-Encode                    x86_64 4:3.21-521.fc44                  fedora       4.7 MiB
 perl-Errno                     x86_64 0:1.38-523.fc45                  fedora       8.4 KiB
 perl-Error                     noarch 1:0.17030-3.fc44                 fedora      76.8 KiB
 perl-Exporter                  noarch 0:5.79-521.fc44                  fedora      54.3 KiB
 perl-Fcntl                     x86_64 0:1.20-523.fc45                  fedora      48.7 KiB
 perl-File-Basename             noarch 0:2.86-523.fc45                  fedora      14.0 KiB
 perl-File-Copy                 noarch 0:2.41-523.fc45                  fedora      19.7 KiB
 perl-File-Path                 noarch 0:2.18-522.fc44                  fedora      63.5 KiB
 perl-File-Temp                 noarch 1:0.231.200-2.fc44               fedora     163.7 KiB
 perl-File-Which                noarch 0:1.27-15.fc44                   fedora      30.4 KiB
 perl-File-stat                 noarch 0:1.14-523.fc45                  fedora      12.5 KiB
 perl-FileHandle                noarch 0:2.05-523.fc45                  fedora       9.4 KiB
 perl-Getopt-Long               noarch 1:2.58-521.fc44                  fedora     144.5 KiB
 perl-Getopt-Std                noarch 0:1.14-523.fc45                  fedora      11.2 KiB
 perl-Git                       noarch 0:2.53.0-1.fc44                  fedora      64.4 KiB
 perl-HTTP-Tiny                 noarch 0:0.092-2.fc44                   fedora     157.6 KiB
 perl-IO                        x86_64 0:1.55-523.fc45                  fedora     147.3 KiB
 perl-IO-Socket-IP              noarch 0:0.43-522.fc44                  fedora     100.3 KiB
 perl-IO-Socket-SSL             noarch 0:2.098-2.fc44                   fedora     723.5 KiB
 perl-IPC-Open3                 noarch 0:1.24-523.fc45                  fedora      27.7 KiB
 perl-MIME-Base32               noarch 0:1.303-25.fc44                  fedora      30.7 KiB
 perl-MIME-Base64               x86_64 0:3.16-521.fc44                  fedora      41.9 KiB
 perl-Net-SSLeay                x86_64 0:1.94-12.fc44                   fedora       1.3 MiB
 perl-POSIX                     x86_64 0:2.23-523.fc45                  fedora     229.4 KiB
 perl-PathTools                 x86_64 0:3.94-521.fc44                  fedora     179.9 KiB
 perl-Pod-Escapes               noarch 1:1.07-521.fc44                  fedora      24.9 KiB
 perl-Pod-Perldoc               noarch 0:3.28.01-522.fc44               fedora     163.7 KiB
 perl-Pod-Simple                noarch 1:3.47-4.fc44                    fedora     565.3 KiB
 perl-Pod-Usage                 noarch 4:2.05-521.fc44                  fedora      86.3 KiB
 perl-Scalar-List-Utils         x86_64 5:1.70-2.fc44                    fedora     144.8 KiB
 perl-SelectSaver               noarch 0:1.02-523.fc45                  fedora       2.2 KiB
 perl-Socket                    x86_64 4:2.040-3.fc44                   fedora     120.1 KiB
 perl-Storable                  x86_64 1:3.37-522.fc44                  fedora     235.1 KiB
 perl-Symbol                    noarch 0:1.09-523.fc45                  fedora       6.8 KiB
 perl-Term-ANSIColor            noarch 0:5.01-522.fc44                  fedora      97.5 KiB
 perl-Term-Cap                  noarch 0:1.18-521.fc44                  fedora      29.3 KiB
 perl-TermReadKey               x86_64 0:2.38-27.fc44                   fedora      63.9 KiB
 perl-Text-ParseWords           noarch 0:3.31-521.fc44                  fedora      13.6 KiB
 perl-Text-Tabs+Wrap            noarch 0:2024.001-521.fc44              fedora      22.6 KiB
 perl-Time-HiRes                x86_64 4:1.9778-521.fc44                fedora     115.6 KiB
 perl-Time-Local                noarch 2:1.350-521.fc44                 fedora      69.0 KiB
 perl-URI                       noarch 0:5.34-3.fc44                    fedora     268.0 KiB
 perl-base                      noarch 0:2.27-523.fc45                  fedora      12.6 KiB
 perl-constant                  noarch 0:1.33-522.fc44                  fedora      26.2 KiB
 perl-if                        noarch 0:0.61.000-523.fc45              fedora       5.8 KiB
 perl-interpreter               x86_64 4:5.42.1-523.fc45                fedora     118.5 KiB
 perl-lib                       x86_64 0:0.65-523.fc45                  fedora       8.5 KiB
 perl-libnet                    noarch 0:3.15-522.fc44                  fedora     289.4 KiB
 perl-libs                      x86_64 4:5.42.1-523.fc45                fedora      11.6 MiB
 perl-locale                    noarch 0:1.13-523.fc45                  fedora       6.1 KiB
 perl-mro                       x86_64 0:1.29-523.fc45                  fedora      41.5 KiB
 perl-overload                  noarch 0:1.40-523.fc45                  fedora      71.6 KiB
 perl-overloading               noarch 0:0.02-523.fc45                  fedora       4.9 KiB
 perl-parent                    noarch 1:0.244-521.fc44                 fedora      10.3 KiB
 perl-podlators                 noarch 1:6.0.2-521.fc44                 fedora     317.5 KiB
 perl-vars                      noarch 0:1.05-523.fc45                  fedora       3.9 KiB
 pmix                           x86_64 0:5.0.7-3.fc44                   fedora       2.2 MiB
 procps-ng                      x86_64 0:4.0.6-1.fc44                   fedora       1.1 MiB
 protobuf-c                     x86_64 0:1.5.2-2.fc44                   fedora      53.7 KiB
 prrte                          x86_64 0:3.0.6-9.fc44                   fedora     161.4 KiB
 prrte-libs                     x86_64 0:3.0.6-9.fc44                   fedora       1.7 MiB
 pthreadpool                    x86_64 0:0.0^git20230829.4fe0e1e-9.fc44 fedora     113.4 KiB
 publicsuffix-list              noarch 0:20260116-1.fc44                fedora     342.2 KiB
 python-pip-wheel               noarch 0:26.0.1-2.fc45                  fedora       1.2 MiB
 python3                        x86_64 0:3.14.3-1.fc45                  fedora      28.7 KiB
 python3-libs                   x86_64 0:3.14.3-1.fc45                  fedora      43.7 MiB
 rdma-core-common               noarch 0:61.0-2.fc44                    fedora      21.9 KiB
 rhash                          x86_64 0:1.4.5-4.fc44                   fedora     354.8 KiB
 rocblas                        x86_64 0:7.2.0-5.fc45                   copr_base  982.9 MiB
 rocm-clang                     x86_64 0:22-5.rocm7.2.0.fc45            copr_base  147.2 MiB
 rocm-clang-devel               x86_64 0:22-5.rocm7.2.0.fc45            copr_base   29.4 MiB
 rocm-clang-libs                x86_64 0:22-5.rocm7.2.0.fc45            copr_base   95.7 MiB
 rocm-clang-runtime-devel       x86_64 0:22-5.rocm7.2.0.fc45            copr_base    8.5 MiB
 rocm-comgr                     x86_64 0:22-5.rocm7.2.0.fc45            copr_base  133.9 MiB
 rocm-device-libs               x86_64 0:22-5.rocm7.2.0.fc45            copr_base    3.3 MiB
 rocm-filesystem                noarch 0:7.2.0-6.fc45                   copr_base    0.0   B
 rocm-hip                       x86_64 0:7.2.0-3.fc45                   copr_base   27.2 MiB
 rocm-lld                       x86_64 0:22-5.rocm7.2.0.fc45            copr_base    6.2 MiB
 rocm-llvm                      x86_64 0:22-5.rocm7.2.0.fc45            copr_base   51.9 MiB
 rocm-llvm-devel                x86_64 0:22-5.rocm7.2.0.fc45            copr_base   34.0 MiB
 rocm-llvm-filesystem           x86_64 0:22-5.rocm7.2.0.fc45            copr_base    0.0   B
 rocm-llvm-libs                 x86_64 0:22-5.rocm7.2.0.fc45            copr_base   96.6 MiB
 rocm-llvm-static               x86_64 0:22-5.rocm7.2.0.fc45            copr_base    2.2 GiB
 rocm-runtime                   x86_64 0:7.2.0-3.fc45                   copr_base    3.9 MiB
 rocminfo                       x86_64 0:7.2.0-3.fc45                   copr_base   81.7 KiB
 rocsolver                      x86_64 0:7.2.0-3.fc45                   copr_base    1.1 GiB
 tcl                            x86_64 1:9.0.2-1.fc44                   fedora       4.3 MiB
 tzdata                         noarch 0:2025c-2.fc44                   fedora       1.2 MiB
 ucx                            x86_64 0:1.19.0-2.fc44                  copr_base    2.5 MiB
 unbound-libs                   x86_64 0:1.24.2-10.fc45                 fedora       1.5 MiB
 vim-filesystem                 noarch 2:9.2.148-1.fc45                 fedora      40.0   B
 wget2                          x86_64 0:2.2.1-2.fc44                   fedora       1.1 MiB
 wget2-libs                     x86_64 0:2.2.1-2.fc44                   fedora     369.5 KiB
 zlib-ng-compat-devel           x86_64 0:2.3.3-5.fc45                   fedora     107.0 KiB

Transaction Summary:
 Installing:       201 packages

Total size of inbound packages is 2 GiB. Need to download 2 GiB.
After this operation, 5 GiB extra will be used (install 5 GiB, remove 0 B).
[  1/201] git-0:2.53.0-1.fc44.x86_64    100% |   2.5 MiB/s |  41.0 KiB |  00m00s
[  2/201] langpacks-en-0:4.3-1.fc45.noa 100% | 516.0 KiB/s |   8.8 KiB |  00m00s
[  3/201] xxd-2:9.2.148-1.fc45.x86_64   100% |   7.7 MiB/s |  31.6 KiB |  00m00s
[  4/201] openmpi-0:5.0.10-1.fc45.x86_6 100% |  78.9 MiB/s |   2.1 MiB |  00m00s
[  5/201] cmake-0:4.2.3-2.fc45.x86_64   100% | 249.4 MiB/s |  14.0 MiB |  00m00s
[  6/201] libcurl-devel-0:8.19.0~rc3-1. 100% | 152.6 MiB/s | 937.4 KiB |  00m00s
[  7/201] gcc-c++-0:16.0.1-0.9.fc45.x86 100% | 113.7 MiB/s |  17.2 MiB |  00m00s
[  8/201] hipblas-devel-0:7.2.0-4.fc45. 100% | 571.0 KiB/s |  82.8 KiB |  00m00s
[  9/201] rocm-comgr-devel-0:22-5.rocm7 100% | 670.4 KiB/s |  32.8 KiB |  00m00s
[ 10/201] rocm-hip-devel-0:7.2.0-3.fc45 100% |  36.9 MiB/s | 264.6 KiB |  00m00s
[ 11/201] rocm-rpm-macros-0:7.2.0-1.fc4 100% |   2.4 MiB/s |  16.9 KiB |  00m00s
[ 12/201] rocm-runtime-devel-0:7.2.0-3. 100% |   6.4 MiB/s | 117.4 KiB |  00m00s
[ 13/201] pthreadpool-devel-0:0.0^git20 100% |  83.3 KiB/s |  14.7 KiB |  00m00s
[ 14/201] git-core-doc-0:2.53.0-1.fc44. 100% | 110.9 MiB/s |   3.1 MiB |  00m00s
[ 15/201] perl-File-Basename-0:2.86-523 100% |   5.5 MiB/s |  16.8 KiB |  00m00s
[ 16/201] perl-Getopt-Long-1:2.58-521.f 100% |  20.7 MiB/s |  63.6 KiB |  00m00s
[ 17/201] perl-Git-0:2.53.0-1.fc44.noar 100% |   4.6 MiB/s |  38.1 KiB |  00m00s
[ 18/201] perl-IPC-Open3-0:1.24-523.fc4 100% |  11.5 MiB/s |  23.6 KiB |  00m00s
[ 19/201] perl-PathTools-0:3.94-521.fc4 100% |  14.2 MiB/s |  87.3 KiB |  00m00s
[ 20/201] perl-TermReadKey-0:2.38-27.fc 100% |  17.3 MiB/s |  35.5 KiB |  00m00s
[ 21/201] perl-interpreter-4:5.42.1-523 100% |   8.8 MiB/s |  71.9 KiB |  00m00s
[ 22/201] perl-lib-0:0.65-523.fc45.x86_ 100% |   4.7 MiB/s |  14.6 KiB |  00m00s
[ 23/201] langpacks-core-en-0:4.3-1.fc4 100% | 462.3 KiB/s |   8.8 KiB |  00m00s
[ 24/201] langpacks-fonts-en-0:4.3-1.fc 100% | 759.0 KiB/s |   9.1 KiB |  00m00s
[ 25/201] rocblas-devel-0:7.2.0-5.fc45. 100% | 521.3 KiB/s | 102.7 KiB |  00m00s
[ 26/201] hwloc-libs-0:2.12.0-3.fc44.x8 100% |  19.8 MiB/s |   2.1 MiB |  00m00s
[ 27/201] libfabric-0:2.3.1-3.fc45.x86_ 100% |  18.9 MiB/s |   1.9 MiB |  00m00s
[ 28/201] libpsm2-0:12.0.1-7.fc45.x86_6 100% |  33.6 MiB/s | 206.3 KiB |  00m00s
[ 29/201] openssh-clients-0:10.2p1-5.fc 100% |  62.8 MiB/s | 771.8 KiB |  00m00s
[ 30/201] git-core-0:2.53.0-1.fc44.x86_ 100% |  22.7 MiB/s |   5.3 MiB |  00m00s
[ 31/201] pmix-0:5.0.7-3.fc44.x86_64    100% |  48.7 MiB/s | 748.3 KiB |  00m00s
[ 32/201] prrte-0:3.0.6-9.fc44.x86_64   100% |   4.6 MiB/s |  56.2 KiB |  00m00s
[ 33/201] cmake-filesystem-0:4.2.3-2.fc 100% |   1.6 MiB/s |  14.9 KiB |  00m00s
[ 34/201] expat-0:2.7.4-1.fc45.x86_64   100% |  25.0 MiB/s | 128.0 KiB |  00m00s
[ 35/201] cmake-data-0:4.2.3-2.fc45.noa 100% |  94.5 MiB/s |   2.6 MiB |  00m00s
[ 36/201] jsoncpp-0:1.9.6-3.fc44.x86_64 100% |  11.9 MiB/s | 110.0 KiB |  00m00s
[ 37/201] libuv-1:1.52.1-1.fc45.x86_64  100% |  53.5 MiB/s | 273.9 KiB |  00m00s
[ 38/201] make-1:4.4.1-12.fc44.x86_64   100% |  71.9 MiB/s | 588.7 KiB |  00m00s
[ 39/201] orangefs-0:2.10.1-3.fc45.x86_ 100% |   9.0 MiB/s | 609.0 KiB |  00m00s
[ 40/201] rhash-0:1.4.5-4.fc44.x86_64   100% |  32.9 MiB/s | 201.9 KiB |  00m00s
[ 41/201] libmpc-0:1.3.1-9.fc44.x86_64  100% |  35.2 MiB/s |  72.1 KiB |  00m00s
[ 42/201] pthreadpool-0:0.0^git20230829 100% | 601.4 KiB/s |  48.7 KiB |  00m00s
[ 43/201] perl-File-Copy-0:2.41-523.fc4 100% |   6.4 MiB/s |  19.8 KiB |  00m00s
[ 44/201] perl-File-Which-0:1.27-15.fc4 100% |   7.0 MiB/s |  21.5 KiB |  00m00s
[ 45/201] perl-Getopt-Std-0:1.14-523.fc 100% | 333.2 KiB/s |  15.3 KiB |  00m00s
[ 46/201] perl-Scalar-List-Utils-5:1.70 100% |  18.4 MiB/s |  75.4 KiB |  00m00s
[ 47/201] perl-URI-0:5.34-3.fc44.noarch 100% |  29.1 MiB/s | 149.2 KiB |  00m00s
[ 48/201] rocm-hip-0:7.2.0-3.fc45.x86_6 100% | 158.2 MiB/s |  10.4 MiB |  00m00s
[ 49/201] environment-modules-0:5.6.1-3 100% |  25.2 MiB/s | 801.3 KiB |  00m00s
[ 50/201] rocm-runtime-0:7.2.0-3.fc45.x 100% |  62.3 MiB/s | 701.9 KiB |  00m00s
[ 51/201] less-0:692-3.fc45.x86_64      100% |  24.0 MiB/s | 221.4 KiB |  00m00s
[ 52/201] perl-Carp-0:1.54-521.fc44.noa 100% |   9.4 MiB/s |  28.8 KiB |  00m00s
[ 53/201] perl-Exporter-0:5.79-521.fc44 100% | 604.3 KiB/s |  30.8 KiB |  00m00s
[ 54/201] perl-Pod-Usage-4:2.05-521.fc4 100% |   9.9 MiB/s |  40.7 KiB |  00m00s
[ 55/201] perl-Text-ParseWords-0:3.31-5 100% |   5.4 MiB/s |  16.5 KiB |  00m00s
[ 56/201] perl-base-0:2.27-523.fc45.noa 100% | 273.2 KiB/s |  15.8 KiB |  00m00s
[ 57/201] perl-constant-0:1.33-522.fc44 100% |  11.2 MiB/s |  22.9 KiB |  00m00s
[ 58/201] perl-overload-0:1.40-523.fc45 100% | 941.5 KiB/s |  45.2 KiB |  00m00s
[ 59/201] perl-Error-1:0.17030-3.fc44.n 100% |  19.7 MiB/s |  40.3 KiB |  00m00s
[ 60/201] perl-Fcntl-0:1.20-523.fc45.x8 100% | 624.9 KiB/s |  29.4 KiB |  00m00s
[ 61/201] perl-IO-0:1.55-523.fc45.x86_6 100% |  26.6 MiB/s |  81.8 KiB |  00m00s
[ 62/201] perl-POSIX-0:2.23-523.fc45.x8 100% |  23.6 MiB/s |  96.5 KiB |  00m00s
[ 63/201] perl-Symbol-0:1.09-523.fc45.n 100% |   6.8 MiB/s |  13.8 KiB |  00m00s
[ 64/201] hipblas-0:7.2.0-4.fc45.x86_64 100% | 228.6 KiB/s | 118.9 KiB |  00m01s
[ 65/201] perl-Errno-0:1.38-523.fc45.x8 100% | 310.2 KiB/s |  14.6 KiB |  00m00s
[ 66/201] perl-DynaLoader-0:1.57-523.fc 100% |   4.2 MiB/s |  25.6 KiB |  00m00s
[ 67/201] perl-libs-4:5.42.1-523.fc45.x 100% |  58.6 MiB/s |   2.6 MiB |  00m00s
[ 68/201] perl-vars-0:1.05-523.fc45.noa 100% |   1.2 MiB/s |  12.6 KiB |  00m00s
[ 69/201] default-fonts-core-sans-0:4.3 100% |  14.2 MiB/s |  29.2 KiB |  00m00s
[ 70/201] google-noto-sans-mono-vf-font 100% |   5.2 MiB/s | 277.3 KiB |  00m00s
[ 71/201] google-noto-serif-vf-fonts-0: 100% |   6.9 MiB/s | 665.8 KiB |  00m00s
[ 72/201] libibverbs-0:61.0-2.fc44.x86_ 100% |  10.5 MiB/s | 492.7 KiB |  00m00s
[ 73/201] libnl3-0:3.12.0-3.fc44.x86_64 100% |  52.6 MiB/s | 377.3 KiB |  00m00s
[ 74/201] librdmacm-0:61.0-2.fc44.x86_6 100% |   6.3 MiB/s |  78.0 KiB |  00m00s
[ 75/201] numactl-libs-0:2.0.19-4.fc44. 100% |   4.5 MiB/s |  32.2 KiB |  00m00s
[ 76/201] libedit-0:3.1-58.20251016cvs. 100% |  35.7 MiB/s | 109.7 KiB |  00m00s
[ 77/201] openssh-0:10.2p1-5.fc45.x86_6 100% |  49.6 MiB/s | 355.3 KiB |  00m00s
[ 78/201] munge-libs-0:0.5.18-1.fc45.x8 100% |   3.0 MiB/s |  21.6 KiB |  00m00s
[ 79/201] prrte-libs-0:3.0.6-9.fc44.x86 100% |  68.8 MiB/s | 563.3 KiB |  00m00s
[ 80/201] emacs-filesystem-1:30.2-2.fc4 100% |   1.6 MiB/s |   8.0 KiB |  00m00s
[ 81/201] vim-filesystem-2:9.2.148-1.fc 100% |   5.1 MiB/s |  15.7 KiB |  00m00s
[ 82/201] perl-Data-Dumper-0:2.191-522. 100% |  11.0 MiB/s |  56.5 KiB |  00m00s
[ 83/201] perl-MIME-Base32-0:1.303-25.f 100% |   6.7 MiB/s |  20.5 KiB |  00m00s
[ 84/201] perl-MIME-Base64-0:3.16-521.f 100% |   9.7 MiB/s |  29.8 KiB |  00m00s
[ 85/201] perl-libnet-0:3.15-522.fc44.n 100% |  12.5 MiB/s | 128.4 KiB |  00m00s
[ 86/201] perl-parent-1:0.244-521.fc44. 100% |   1.3 MiB/s |  14.9 KiB |  00m00s
[ 87/201] libdrm-0:2.4.131-1.fc44.x86_6 100% |   9.6 MiB/s | 166.3 KiB |  00m00s
[ 88/201] perl-Pod-Perldoc-0:3.28.01-52 100% |  28.0 MiB/s |  86.1 KiB |  00m00s
[ 89/201] perl-podlators-1:6.0.2-521.fc 100% |  31.4 MiB/s | 128.5 KiB |  00m00s
[ 90/201] perl-mro-0:1.29-523.fc45.x86_ 100% |   5.8 MiB/s |  29.5 KiB |  00m00s
[ 91/201] perl-overloading-0:0.02-523.f 100% |   6.1 MiB/s |  12.5 KiB |  00m00s
[ 92/201] perl-File-stat-0:1.14-523.fc4 100% |   2.7 MiB/s |  16.7 KiB |  00m00s
[ 93/201] perl-SelectSaver-0:1.02-523.f 100% |   5.5 MiB/s |  11.4 KiB |  00m00s
[ 94/201] perl-Socket-4:2.040-3.fc44.x8 100% |   6.0 MiB/s |  55.3 KiB |  00m00s
[ 95/201] perl-locale-0:1.13-523.fc45.n 100% |   6.4 MiB/s |  13.1 KiB |  00m00s
[ 96/201] abattis-cantarell-vf-fonts-0: 100% |  14.7 MiB/s | 120.4 KiB |  00m00s
[ 97/201] man-db-0:2.13.1-3.fc44.x86_64 100% |  20.1 MiB/s |   1.4 MiB |  00m00s
[ 98/201] fonts-filesystem-1:5.0.0-3.fc 100% |   1.7 MiB/s |   8.7 KiB |  00m00s
[ 99/201] google-noto-fonts-common-0:20 100% |   8.5 MiB/s |  17.3 KiB |  00m00s
[100/201] rdma-core-common-0:61.0-2.fc4 100% |   2.4 MiB/s |  16.9 KiB |  00m00s
[101/201] google-noto-sans-vf-fonts-0:2 100% |  25.0 MiB/s | 614.5 KiB |  00m00s
[102/201] perl-B-0:1.89-523.fc45.x86_64 100% |  28.9 MiB/s | 177.5 KiB |  00m00s
[103/201] perl-Digest-MD5-0:2.59-521.fc 100% |  17.6 MiB/s |  36.0 KiB |  00m00s
[104/201] perl-FileHandle-0:2.05-523.fc 100% |   4.9 MiB/s |  15.1 KiB |  00m00s
[105/201] perl-IO-Socket-IP-0:0.43-522. 100% |  10.3 MiB/s |  42.2 KiB |  00m00s
[106/201] perl-Time-Local-2:1.350-521.f 100% |   2.8 MiB/s |  34.5 KiB |  00m00s
[107/201] groff-base-0:1.24.0-1.fc45.x8 100% |  55.8 MiB/s |   1.2 MiB |  00m00s
[108/201] libpipeline-0:1.5.8-4.fc44.x8 100% |   3.7 MiB/s |  60.3 KiB |  00m00s
[109/201] libpciaccess-0:0.16-17.fc44.x 100% |   8.9 MiB/s |  27.2 KiB |  00m00s
[110/201] perl-File-Temp-1:0.231.200-2. 100% |  11.6 MiB/s |  59.6 KiB |  00m00s
[111/201] perl-HTTP-Tiny-0:0.092-2.fc44 100% |   5.1 MiB/s |  57.2 KiB |  00m00s
[112/201] perl-Pod-Simple-1:3.47-4.fc44 100% |  30.7 MiB/s | 220.0 KiB |  00m00s
[113/201] perl-Term-ANSIColor-0:5.01-52 100% |  15.5 MiB/s |  47.7 KiB |  00m00s
[114/201] perl-Term-Cap-0:1.18-521.fc44 100% |   7.2 MiB/s |  22.0 KiB |  00m00s
[115/201] perl-Class-Struct-0:0.68-523. 100% |   7.1 MiB/s |  21.7 KiB |  00m00s
[116/201] perl-if-0:0.61.000-523.fc45.n 100% |   3.3 MiB/s |  13.6 KiB |  00m00s
[117/201] perl-Digest-0:1.20-521.fc44.n 100% |   4.0 MiB/s |  24.9 KiB |  00m00s
[118/201] perl-File-Path-0:2.18-522.fc4 100% |   4.9 MiB/s |  35.3 KiB |  00m00s
[119/201] hwdata-0:0.405-1.fc45.noarch  100% | 120.4 MiB/s |   1.7 MiB |  00m00s
[120/201] perl-IO-Socket-SSL-0:2.098-2. 100% |  57.3 MiB/s | 234.7 KiB |  00m00s
[121/201] perl-Net-SSLeay-0:1.94-12.fc4 100% |  91.2 MiB/s | 373.5 KiB |  00m00s
[122/201] perl-Time-HiRes-4:1.9778-521. 100% |  14.0 MiB/s |  57.2 KiB |  00m00s
[123/201] perl-Pod-Escapes-1:1.07-521.f 100% |   6.4 MiB/s |  19.8 KiB |  00m00s
[124/201] perl-Text-Tabs+Wrap-0:2024.00 100% |   4.2 MiB/s |  21.8 KiB |  00m00s
[125/201] ncurses-0:6.6-1.fc44.x86_64   100% |  46.6 MiB/s | 429.8 KiB |  00m00s
[126/201] perl-AutoLoader-0:5.74-523.fc 100% |   1.5 MiB/s |  20.9 KiB |  00m00s
[127/201] wget2-wget-0:2.2.1-2.fc44.x86 100% | 291.8 KiB/s |   9.3 KiB |  00m00s
[128/201] wget2-0:2.2.1-2.fc44.x86_64   100% |  10.2 MiB/s | 292.5 KiB |  00m00s
[129/201] gpgme-0:2.0.1-3.fc44.x86_64   100% |  28.3 MiB/s | 231.7 KiB |  00m00s
[130/201] wget2-libs-0:2.2.1-2.fc44.x86 100% |  29.6 MiB/s | 151.7 KiB |  00m00s
[131/201] gnutls-dane-0:3.8.12-1.fc45.x 100% |  13.0 MiB/s |  39.9 KiB |  00m00s
[132/201] unbound-libs-0:1.24.2-10.fc45 100% |  73.3 MiB/s | 600.7 KiB |  00m00s
[133/201] dns-root-data-0:2026260100-2. 100% |   1.6 MiB/s |  14.6 KiB |  00m00s
[134/201] hiredis-0:1.2.0-8.fc44.x86_64 100% |  17.1 MiB/s |  52.6 KiB |  00m00s
[135/201] protobuf-c-0:1.5.2-2.fc44.x86 100% | 543.5 KiB/s |  33.7 KiB |  00m00s
[136/201] mpdecimal-0:4.0.1-3.fc44.x86_ 100% |  24.2 MiB/s |  99.1 KiB |  00m00s
[137/201] python-pip-wheel-0:26.0.1-2.f 100% |  88.4 MiB/s |   1.1 MiB |  00m00s
[138/201] tzdata-0:2025c-2.fc44.noarch  100% |  99.6 MiB/s | 714.2 KiB |  00m00s
[139/201] hipcc-0:22-5.rocm7.2.0.fc45.x 100% |  37.0 MiB/s | 151.4 KiB |  00m00s
[140/201] python3-libs-0:3.14.3-1.fc45. 100% |  95.3 MiB/s |  10.1 MiB |  00m00s
[141/201] rocm-device-libs-0:22-5.rocm7 100% |  40.6 MiB/s | 498.4 KiB |  00m00s
[142/201] rocminfo-0:7.2.0-3.fc45.x86_6 100% |   3.6 MiB/s |  40.8 KiB |  00m00s
[143/201] kmod-0:34.2-4.fc44.x86_64     100% |  11.1 MiB/s | 136.1 KiB |  00m00s
[144/201] rocm-comgr-0:22-5.rocm7.2.0.f 100% | 295.8 MiB/s |  32.2 MiB |  00m00s
[145/201] hipblas-common-devel-0:7.2.0- 100% |  74.5 KiB/s |  13.6 KiB |  00m00s
[146/201] gcc-0:16.0.1-0.9.fc45.x86_64  100% | 204.4 MiB/s |  42.9 MiB |  00m00s
[147/201] libstdc++-devel-0:16.0.1-0.9. 100% | 132.1 MiB/s |   5.4 MiB |  00m00s
[148/201] cpp-0:16.0.1-0.9.fc45.x86_64  100% | 162.4 MiB/s |  14.5 MiB |  00m00s
[149/201] glibc-devel-0:2.43.9000-4.fc4 100% |  86.6 MiB/s | 620.9 KiB |  00m00s
[150/201] libxcrypt-devel-0:4.5.2-3.fc4 100% |   9.8 MiB/s |  30.0 KiB |  00m00s
[151/201] libatomic-0:16.0.1-0.9.fc45.x 100% |   8.4 MiB/s |  42.8 KiB |  00m00s
[152/201] perl-Encode-4:3.21-521.fc44.x 100% |  21.0 MiB/s |   1.1 MiB |  00m00s
[153/201] perl-Storable-1:3.37-522.fc44 100% |  32.8 MiB/s | 100.9 KiB |  00m00s
[154/201] libgfortran-0:16.0.1-0.9.fc45 100% |   9.7 MiB/s | 981.7 KiB |  00m00s
[155/201] libquadmath-0:16.0.1-0.9.fc45 100% |   6.1 MiB/s | 194.4 KiB |  00m00s
[156/201] ucx-0:1.19.0-2.fc44.x86_64    100% |   3.0 MiB/s | 898.0 KiB |  00m00s
[157/201] brotli-devel-0:1.2.0-3.fc44.x 100% |  11.3 MiB/s |  34.6 KiB |  00m00s
[158/201] brotli-0:1.2.0-3.fc44.x86_64  100% | 389.1 KiB/s |  24.1 KiB |  00m00s
[159/201] krb5-devel-0:1.22.2-2.fc45.x8 100% |  34.9 MiB/s | 143.0 KiB |  00m00s
[160/201] libkadm5-0:1.22.2-2.fc45.x86_ 100% |  26.0 MiB/s |  79.7 KiB |  00m00s
[161/201] libidn2-devel-0:2.3.8-3.fc44. 100% |  20.9 MiB/s |  64.1 KiB |  00m00s
[162/201] libnghttp2-devel-0:1.68.0-3.f 100% |  26.7 MiB/s |  54.6 KiB |  00m00s
[163/201] libnghttp3-devel-0:1.15.0-1.f 100% | 471.0 KiB/s |  26.4 KiB |  00m00s
[164/201] libpsl-devel-0:0.21.5-7.fc44. 100% |  10.7 MiB/s |  32.8 KiB |  00m00s
[165/201] publicsuffix-list-0:20260116- 100% |   2.0 MiB/s |  91.7 KiB |  00m00s
[166/201] libssh-devel-0:0.12.0-1.fc45. 100% |  23.2 MiB/s |  47.6 KiB |  00m00s
[167/201] ngtcp2-crypto-ossl-devel-0:1. 100% | 197.0 KiB/s |  11.0 KiB |  00m00s
[168/201] ngtcp2-devel-0:1.19.0-2.fc44. 100% |  25.9 MiB/s |  53.0 KiB |  00m00s
[169/201] openssl-devel-1:3.5.5-1.fc44. 100% |  48.7 MiB/s |   3.0 MiB |  00m00s
[170/201] zlib-ng-compat-devel-0:2.3.3- 100% |  18.7 MiB/s |  38.2 KiB |  00m00s
[171/201] keyutils-libs-devel-0:1.6.3-7 100% |   1.1 MiB/s |  59.7 KiB |  00m00s
[172/201] libcom_err-devel-0:1.47.4-1.f 100% |   8.2 MiB/s |  16.8 KiB |  00m00s
[173/201] libselinux-devel-0:3.10-1.fc4 100% |   2.3 MiB/s | 151.7 KiB |  00m00s
[174/201] libsepol-devel-0:3.10-1.fc44. 100% |  23.5 MiB/s |  48.2 KiB |  00m00s
[175/201] libverto-devel-0:0.3.2-12.fc4 100% | 319.9 KiB/s |  14.4 KiB |  00m00s
[176/201] kernel-headers-0:7.0.0-0.rc4. 100% | 133.8 MiB/s |   1.7 MiB |  00m00s
[177/201] procps-ng-0:4.0.6-1.fc44.x86_ 100% |  77.9 MiB/s | 398.9 KiB |  00m00s
[178/201] tcl-1:9.0.2-1.fc44.x86_64     100% | 112.1 MiB/s |   1.2 MiB |  00m00s
[179/201] libtommath-0:1.3.1~rc1-7.fc44 100% |  12.7 MiB/s |  64.8 KiB |  00m00s
[180/201] python3-0:3.14.3-1.fc45.x86_6 100% |   9.2 MiB/s |  28.2 KiB |  00m00s
[181/201] rocm-clang-devel-0:22-5.rocm7 100% | 179.1 MiB/s |   2.9 MiB |  00m00s
[182/201] rocm-lld-0:22-5.rocm7.2.0.fc4 100% | 102.1 MiB/s |   1.6 MiB |  00m00s
[183/201] rocm-llvm-static-0:22-5.rocm7 100% | 188.7 MiB/s | 303.8 MiB |  00m02s
[184/201] rocm-clang-0:22-5.rocm7.2.0.f 100% | 266.6 MiB/s |  35.2 MiB |  00m00s
[185/201] rocm-clang-libs-0:22-5.rocm7. 100% | 273.9 MiB/s |  23.8 MiB |  00m00s
[186/201] rocm-clang-runtime-devel-0:22 100% |  70.2 MiB/s | 646.8 KiB |  00m00s
[187/201] rocm-llvm-libs-0:22-5.rocm7.2 100% | 110.0 MiB/s |  22.2 MiB |  00m00s
[188/201] pcre2-devel-0:10.47-1.fc44.1. 100% | 107.6 MiB/s | 550.9 KiB |  00m00s
[189/201] pcre2-utf16-0:10.47-1.fc44.1. 100% |  83.4 MiB/s | 256.3 KiB |  00m00s
[190/201] pcre2-utf32-0:10.47-1.fc44.1. 100% |  59.1 MiB/s | 242.2 KiB |  00m00s
[191/201] rocm-llvm-devel-0:22-5.rocm7. 100% | 258.5 MiB/s |   4.7 MiB |  00m00s
[192/201] rocm-llvm-filesystem-0:22-5.r 100% |   5.7 MiB/s |  17.5 KiB |  00m00s
[193/201] rocm-filesystem-0:7.2.0-6.fc4 100% |   1.6 MiB/s |  15.1 KiB |  00m00s
[194/201] rocm-llvm-0:22-5.rocm7.2.0.fc 100% |  88.3 MiB/s |  14.0 MiB |  00m00s
[195/201] annobin-plugin-gcc-0:13.07-1. 100% | 133.6 MiB/s | 683.8 KiB |  00m00s
[196/201] annobin-docs-0:13.07-1.fc45.n 100% |  21.6 MiB/s |  88.3 KiB |  00m00s
[197/201] cmake-rpm-macros-0:4.2.3-2.fc 100% |   6.8 MiB/s |  13.9 KiB |  00m00s
[198/201] ninja-build-0:1.13.2-2.fc44.x 100% |   4.3 MiB/s | 203.9 KiB |  00m00s
[199/201] gcc-plugin-annobin-0:16.0.1-0 100% |   9.1 MiB/s |  46.4 KiB |  00m00s
[200/201] rocblas-0:7.2.0-5.fc45.x86_64 100% |  35.8 MiB/s | 299.0 MiB |  00m08s
[201/201] rocsolver-0:7.2.0-3.fc45.x86_ 100% |  81.4 MiB/s |   1.0 GiB |  00m12s
--------------------------------------------------------------------------------
[201/201] Total                         100% | 136.1 MiB/s |   1.9 GiB |  00m14s
Running transaction
[  1/203] Verify package files          100% |  23.0   B/s | 201.0   B |  00m08s
[  2/203] Prepare transaction           100% |   1.2 KiB/s | 201.0   B |  00m00s
[  3/203] Installing cmake-filesystem-0 100% |   4.4 MiB/s |   9.1 KiB |  00m00s
[  4/203] Installing fonts-filesystem-1 100% |   0.0   B/s | 788.0   B |  00m00s
[  5/203] Installing hwloc-libs-0:2.12. 100% | 414.5 MiB/s |   2.9 MiB |  00m00s
[  6/203] Installing google-noto-fonts- 100% |   0.0   B/s |  18.5 KiB |  00m00s
[  7/203] Installing vim-filesystem-2:9 100% |   4.7 MiB/s |   4.8 KiB |  00m00s
[  8/203] Installing numactl-libs-0:2.0 100% |  56.3 MiB/s |  57.7 KiB |  00m00s
[  9/203] Installing libnl3-0:3.12.0-3. 100% | 266.6 MiB/s |   1.1 MiB |  00m00s
[ 10/203] Installing less-0:692-3.fc45. 100% |  22.9 MiB/s | 468.9 KiB |  00m00s
[ 11/203] Installing libmpc-0:1.3.1-9.f 100% | 158.2 MiB/s | 162.0 KiB |  00m00s
[ 12/203] Installing expat-0:2.7.4-1.fc 100% |  16.6 MiB/s | 323.3 KiB |  00m00s
[ 13/203] Installing libpsm2-0:12.0.1-7 100% | 222.3 MiB/s | 455.3 KiB |  00m00s
[ 14/203] Installing zlib-ng-compat-dev 100% |  13.3 MiB/s | 108.6 KiB |  00m00s
[ 15/203] Installing openssl-devel-1:3. 100% |  61.0 MiB/s |   5.4 MiB |  00m00s
[ 16/203] Installing ngtcp2-devel-0:1.1 100% | 277.0 MiB/s | 283.6 KiB |  00m00s
[ 17/203] Installing rocm-comgr-0:22-5. 100% |  68.2 MiB/s | 133.9 MiB |  00m02s
[ 18/203] Installing gpgme-0:2.0.1-3.fc 100% |  21.7 MiB/s | 620.8 KiB |  00m00s
[ 19/203] Installing groff-base-0:1.24. 100% |  98.3 MiB/s |   4.2 MiB |  00m00s
[ 20/203] Installing rdma-core-common-0 100% |  22.6 MiB/s |  23.2 KiB |  00m00s
[ 21/203] Installing libibverbs-0:61.0- 100% | 212.3 MiB/s |   1.3 MiB |  00m00s
[ 22/203] Installing emacs-filesystem-1 100% |   0.0   B/s | 812.0   B |  00m00s
[ 23/203] Installing make-1:4.4.1-12.fc 100% |  75.0 MiB/s |   1.8 MiB |  00m00s
[ 24/203] Installing orangefs-0:2.10.1- 100% |  67.3 MiB/s |   1.7 MiB |  00m00s
[ 25/203] Installing librdmacm-0:61.0-2 100% | 152.1 MiB/s | 155.8 KiB |  00m00s
[ 26/203] Installing libfabric-0:2.3.1- 100% | 231.8 MiB/s |   9.3 MiB |  00m00s
[ 27/203] Installing ngtcp2-crypto-ossl 100% |   7.9 MiB/s |   8.1 KiB |  00m00s
[ 28/203] Installing cpp-0:16.0.1-0.9.f 100% | 304.9 MiB/s |  42.1 MiB |  00m00s
[ 29/203] Installing google-noto-sans-m 100% | 274.5 MiB/s | 562.2 KiB |  00m00s
[ 30/203] Installing google-noto-serif- 100% | 318.0 MiB/s |   1.6 MiB |  00m00s
[ 31/203] Installing google-noto-sans-v 100% | 278.3 MiB/s |   1.4 MiB |  00m00s
[ 32/203] Installing abattis-cantarell- 100% | 189.9 MiB/s | 194.4 KiB |  00m00s
[ 33/203] Installing default-fonts-core 100% |  17.8 MiB/s |  18.2 KiB |  00m00s
[ 34/203] Installing langpacks-core-en- 100% |   0.0   B/s | 704.0   B |  00m00s
[ 35/203] Installing langpacks-fonts-en 100% |   0.0   B/s | 652.0   B |  00m00s
[ 36/203] Installing hipblas-common-dev 100% |   0.0   B/s |  18.2 KiB |  00m00s
[ 37/203] Installing annobin-docs-0:13. 100% |  98.0 MiB/s | 100.3 KiB |  00m00s
[ 38/203] Installing rocm-filesystem-0: 100% |   7.7 MiB/s |  15.8 KiB |  00m00s
[ 39/203] Installing rocm-llvm-filesyst 100% |   3.5 MiB/s |   3.6 KiB |  00m00s
[ 40/203] Installing rocm-llvm-libs-0:2 100% |  70.4 MiB/s |  96.6 MiB |  00m01s
[ 41/203] Installing rocm-clang-libs-0: 100% |  68.4 MiB/s |  95.7 MiB |  00m01s
[ 42/203] Installing rocm-lld-0:22-5.ro 100% |  64.9 MiB/s |   6.2 MiB |  00m00s
[ 43/203] Installing pcre2-utf32-0:10.4 100% | 304.7 MiB/s | 623.9 KiB |  00m00s
[ 44/203] Installing pcre2-utf16-0:10.4 100% | 213.6 MiB/s | 656.1 KiB |  00m00s
[ 45/203] Installing pcre2-devel-0:10.4 100% |  81.5 MiB/s |   2.1 MiB |  00m00s
[ 46/203] Installing rocm-clang-runtime 100% | 126.0 MiB/s |   8.6 MiB |  00m00s
[ 47/203] Installing libtommath-0:1.3.1 100% | 128.3 MiB/s | 131.4 KiB |  00m00s
[ 48/203] Installing tcl-1:9.0.2-1.fc44 100% | 127.6 MiB/s |   4.3 MiB |  00m00s
[ 49/203] Installing procps-ng-0:4.0.6- 100% |  44.0 MiB/s |   1.1 MiB |  00m00s
[ 50/203] Installing kernel-headers-0:7 100% | 186.1 MiB/s |   7.1 MiB |  00m00s
[ 51/203] Installing glibc-devel-0:2.43 100% | 159.1 MiB/s |   2.4 MiB |  00m00s
[ 52/203] Installing libxcrypt-devel-0: 100% |  32.5 MiB/s |  33.3 KiB |  00m00s
[ 53/203] Installing libverto-devel-0:0 100% |   0.0   B/s |  26.4 KiB |  00m00s
[ 54/203] Installing libsepol-devel-0:3 100% |  63.0 MiB/s | 129.0 KiB |  00m00s
[ 55/203] Installing libselinux-devel-0 100% |  38.3 MiB/s | 156.9 KiB |  00m00s
[ 56/203] Installing libcom_err-devel-0 100% | 961.8 KiB/s |  18.3 KiB |  00m00s
[ 57/203] Installing keyutils-libs-deve 100% |  53.9 MiB/s |  55.2 KiB |  00m00s
[ 58/203] Installing publicsuffix-list- 100% | 335.2 MiB/s | 343.2 KiB |  00m00s
[ 59/203] Installing libpsl-devel-0:0.2 100% | 110.9 MiB/s | 113.5 KiB |  00m00s
[ 60/203] Installing libnghttp3-devel-0 100% | 105.7 MiB/s | 108.2 KiB |  00m00s
[ 61/203] Installing libnghttp2-devel-0 100% | 282.3 MiB/s | 289.1 KiB |  00m00s
[ 62/203] Installing libidn2-devel-0:2. 100% |  76.5 MiB/s | 156.7 KiB |  00m00s
[ 63/203] Installing libkadm5-0:1.22.2- 100% | 214.6 MiB/s | 219.8 KiB |  00m00s
[ 64/203] Installing krb5-devel-0:1.22. 100% |  35.1 MiB/s | 719.2 KiB |  00m00s
[ 65/203] Installing libssh-devel-0:0.1 100% | 201.4 MiB/s | 206.2 KiB |  00m00s
[ 66/203] Installing brotli-0:1.2.0-3.f 100% |   1.9 MiB/s |  34.3 KiB |  00m00s
[ 67/203] Installing brotli-devel-0:1.2 100% |  66.8 MiB/s |  68.4 KiB |  00m00s
[ 68/203] Installing ucx-0:1.19.0-2.fc4 100% |  96.7 MiB/s |   2.5 MiB |  00m00s
[ 69/203] Installing libquadmath-0:16.0 100% | 159.7 MiB/s | 327.1 KiB |  00m00s
[ 70/203] Installing libgfortran-0:16.0 100% | 378.9 MiB/s |   3.4 MiB |  00m00s
[ 71/203] Installing libatomic-0:16.0.1 100% |  45.5 MiB/s |  46.6 KiB |  00m00s
[ 72/203] Installing gcc-0:16.0.1-0.9.f 100% | 356.4 MiB/s | 120.4 MiB |  00m00s
[ 73/203] Installing libstdc++-devel-0: 100% | 419.2 MiB/s |  39.0 MiB |  00m00s
[ 74/203] Installing kmod-0:34.2-4.fc44 100% |  10.0 MiB/s | 256.9 KiB |  00m00s
[ 75/203] Installing tzdata-0:2025c-2.f 100% |  38.8 MiB/s |   1.5 MiB |  00m00s
[ 76/203] Installing python-pip-wheel-0 100% | 611.4 MiB/s |   1.2 MiB |  00m00s
[ 77/203] Installing mpdecimal-0:4.0.1- 100% |  30.5 MiB/s | 218.6 KiB |  00m00s
[ 78/203] Installing python3-libs-0:3.1 100% | 306.4 MiB/s |  44.1 MiB |  00m00s
[ 79/203] Installing python3-0:3.14.3-1 100% |   1.5 MiB/s |  30.5 KiB |  00m00s
[ 80/203] Installing rocm-llvm-0:22-5.r 100% |  62.6 MiB/s |  51.9 MiB |  00m01s
[ 81/203] Installing rocm-llvm-devel-0: 100% |  91.2 MiB/s |  34.4 MiB |  00m00s
[ 82/203] Installing rocm-llvm-static-0 100% |  92.0 MiB/s |   2.2 GiB |  00m25s
[ 83/203] Installing ninja-build-0:1.13 100% |  23.2 MiB/s | 499.6 KiB |  00m00s
[ 84/203] Installing cmake-rpm-macros-0 100% |   8.8 MiB/s |   9.0 KiB |  00m00s
[ 85/203] Installing protobuf-c-0:1.5.2 100% |  53.9 MiB/s |  55.2 KiB |  00m00s
[ 86/203] Installing hiredis-0:1.2.0-8. 100% | 104.9 MiB/s | 107.4 KiB |  00m00s
[ 87/203] Installing dns-root-data-0:20 100% |   1.5 MiB/s |  12.5 KiB |  00m00s
>>> Running sysusers scriptlet: unbound-libs-0:1.24.2-10.fc45.x86_64
>>> Finished sysusers scriptlet: unbound-libs-0:1.24.2-10.fc45.x86_64
>>> Scriptlet output:
>>> Creating group 'unbound' with GID 999.
>>> Creating user 'unbound' (Unbound DNS resolver) with UID 999 and GID 999.
>>> 
[ 88/203] Installing unbound-libs-0:1.2 100% | 255.1 MiB/s |   1.5 MiB |  00m00s
[ 89/203] Installing gnutls-dane-0:3.8. 100% |  60.2 MiB/s |  61.6 KiB |  00m00s
[ 90/203] Installing wget2-libs-0:2.2.1 100% | 181.0 MiB/s | 370.7 KiB |  00m00s
[ 91/203] Installing wget2-0:2.2.1-2.fc 100% |  47.6 MiB/s |   1.1 MiB |  00m00s
[ 92/203] Installing ncurses-0:6.6-1.fc 100% |  28.7 MiB/s | 618.0 KiB |  00m00s
[ 93/203] Installing perl-Digest-0:1.20 100% |  36.2 MiB/s |  37.1 KiB |  00m00s
[ 94/203] Installing perl-Digest-MD5-0: 100% |  60.0 MiB/s |  61.5 KiB |  00m00s
[ 95/203] Installing perl-B-0:1.89-523. 100% | 246.4 MiB/s | 504.6 KiB |  00m00s
[ 96/203] Installing perl-FileHandle-0: 100% |   0.0   B/s |   9.8 KiB |  00m00s
[ 97/203] Installing perl-libnet-0:3.15 100% | 143.9 MiB/s | 294.7 KiB |  00m00s
[ 98/203] Installing perl-Data-Dumper-0 100% | 114.7 MiB/s | 117.4 KiB |  00m00s
[ 99/203] Installing perl-MIME-Base32-0 100% |   0.0   B/s |  32.2 KiB |  00m00s
[100/203] Installing perl-AutoLoader-0: 100% |  20.5 MiB/s |  21.0 KiB |  00m00s
[101/203] Installing perl-URI-0:5.34-3. 100% |  91.7 MiB/s | 281.8 KiB |  00m00s
[102/203] Installing perl-IO-Socket-IP- 100% |  99.8 MiB/s | 102.2 KiB |  00m00s
[103/203] Installing perl-Net-SSLeay-0: 100% | 224.5 MiB/s |   1.3 MiB |  00m00s
[104/203] Installing perl-IO-Socket-SSL 100% | 236.9 MiB/s | 727.6 KiB |  00m00s
[105/203] Installing perl-Text-Tabs+Wra 100% |   0.0   B/s |  23.9 KiB |  00m00s
[106/203] Installing perl-Pod-Escapes-1 100% |   0.0   B/s |  25.9 KiB |  00m00s
[107/203] Installing perl-Time-HiRes-4: 100% | 114.9 MiB/s | 117.7 KiB |  00m00s
[108/203] Installing perl-File-Path-0:2 100% |   0.0   B/s |  64.5 KiB |  00m00s
[109/203] Installing perl-if-0:0.61.000 100% |   0.0   B/s |   6.2 KiB |  00m00s
[110/203] Installing perl-Time-Local-2: 100% |  68.9 MiB/s |  70.6 KiB |  00m00s
[111/203] Installing perl-locale-0:1.13 100% |   0.0   B/s |   6.5 KiB |  00m00s
[112/203] Installing perl-Pod-Simple-1: 100% | 187.1 MiB/s | 574.9 KiB |  00m00s
[113/203] Installing perl-HTTP-Tiny-0:0 100% | 155.9 MiB/s | 159.7 KiB |  00m00s
[114/203] Installing perl-File-Temp-1:0 100% | 161.6 MiB/s | 165.5 KiB |  00m00s
[115/203] Installing perl-Class-Struct- 100% |   0.0   B/s |  25.9 KiB |  00m00s
[116/203] Installing perl-IPC-Open3-0:1 100% |   0.0   B/s |  28.5 KiB |  00m00s
[117/203] Installing perl-Term-Cap-0:1. 100% |   0.0   B/s |  30.6 KiB |  00m00s
[118/203] Installing perl-Term-ANSIColo 100% |  96.9 MiB/s |  99.2 KiB |  00m00s
[119/203] Installing perl-POSIX-0:2.23- 100% | 225.2 MiB/s | 230.6 KiB |  00m00s
[120/203] Installing perl-podlators-1:6 100% |  15.7 MiB/s | 321.4 KiB |  00m00s
[121/203] Installing perl-Pod-Perldoc-0 100% |   8.7 MiB/s | 169.2 KiB |  00m00s
[122/203] Installing perl-File-stat-0:1 100% |  12.8 MiB/s |  13.1 KiB |  00m00s
[123/203] Installing perl-Socket-4:2.04 100% | 119.3 MiB/s | 122.2 KiB |  00m00s
[124/203] Installing perl-SelectSaver-0 100% |   0.0   B/s |   2.6 KiB |  00m00s
[125/203] Installing perl-Symbol-0:1.09 100% |   0.0   B/s |   7.3 KiB |  00m00s
[126/203] Installing perl-Pod-Usage-4:2 100% |   4.8 MiB/s |  87.9 KiB |  00m00s
[127/203] Installing perl-IO-0:1.55-523 100% | 148.0 MiB/s | 151.5 KiB |  00m00s
[128/203] Installing perl-overloading-0 100% |   0.0   B/s |   5.6 KiB |  00m00s
[129/203] Installing perl-mro-0:1.29-52 100% |  41.6 MiB/s |  42.6 KiB |  00m00s
[130/203] Installing perl-Fcntl-0:1.20- 100% |   0.0   B/s |  49.8 KiB |  00m00s
[131/203] Installing perl-base-0:2.27-5 100% |   0.0   B/s |  13.0 KiB |  00m00s
[132/203] Installing perl-Text-ParseWor 100% |   0.0   B/s |  14.6 KiB |  00m00s
[133/203] Installing perl-File-Basename 100% |   0.0   B/s |  14.6 KiB |  00m00s
[134/203] Installing perl-Getopt-Long-1 100% | 143.8 MiB/s | 147.2 KiB |  00m00s
[135/203] Installing perl-Storable-1:3. 100% | 231.2 MiB/s | 236.7 KiB |  00m00s
[136/203] Installing perl-overload-0:1. 100% |   0.0   B/s |  72.0 KiB |  00m00s
[137/203] Installing perl-parent-1:0.24 100% |   0.0   B/s |  11.0 KiB |  00m00s
[138/203] Installing perl-MIME-Base64-0 100% |  43.1 MiB/s |  44.2 KiB |  00m00s
[139/203] Installing perl-vars-0:1.05-5 100% |   0.0   B/s |   4.3 KiB |  00m00s
[140/203] Installing perl-Errno-0:1.38- 100% |   0.0   B/s |   8.8 KiB |  00m00s
[141/203] Installing perl-constant-0:1. 100% |   0.0   B/s |  27.4 KiB |  00m00s
[142/203] Installing perl-Scalar-List-U 100% | 145.0 MiB/s | 148.5 KiB |  00m00s
[143/203] Installing perl-Getopt-Std-0: 100% |   0.0   B/s |  11.8 KiB |  00m00s
[144/203] Installing perl-Encode-4:3.21 100% | 151.4 MiB/s |   4.7 MiB |  00m00s
[145/203] Installing perl-DynaLoader-0: 100% |  31.7 MiB/s |  32.5 KiB |  00m00s
[146/203] Installing perl-PathTools-0:3 100% | 180.1 MiB/s | 184.4 KiB |  00m00s
[147/203] Installing perl-Exporter-0:5. 100% |   0.0   B/s |  55.6 KiB |  00m00s
[148/203] Installing perl-Carp-0:1.54-5 100% |   6.7 MiB/s |  47.7 KiB |  00m00s
[149/203] Installing perl-libs-4:5.42.1 100% | 271.6 MiB/s |  11.7 MiB |  00m00s
[150/203] Installing perl-interpreter-4 100% |   6.2 MiB/s | 120.2 KiB |  00m00s
[151/203] Installing perl-TermReadKey-0 100% |  64.5 MiB/s |  66.1 KiB |  00m00s
[152/203] Installing perl-lib-0:0.65-52 100% |   0.0   B/s |   8.9 KiB |  00m00s
[153/203] Installing perl-File-Copy-0:2 100% |   0.0   B/s |  20.2 KiB |  00m00s
[154/203] Installing perl-File-Which-0: 100% |   0.0   B/s |  31.4 KiB |  00m00s
[155/203] Installing perl-Error-1:0.170 100% |  78.1 MiB/s |  80.0 KiB |  00m00s
[156/203] Installing hwdata-0:0.405-1.f 100% | 491.6 MiB/s |   9.8 MiB |  00m00s
[157/203] Installing libpciaccess-0:0.1 100% |  48.6 MiB/s |  49.8 KiB |  00m00s
[158/203] Installing libdrm-0:2.4.131-1 100% | 199.2 MiB/s | 408.0 KiB |  00m00s
[159/203] Installing rocm-runtime-0:7.2 100% | 484.5 MiB/s |   3.9 MiB |  00m00s
[160/203] Installing rocm-runtime-devel 100% | 336.0 MiB/s | 688.2 KiB |  00m00s
[161/203] Installing rocminfo-0:7.2.0-3 100% |   4.3 MiB/s |  83.1 KiB |  00m00s
[162/203] Installing libpipeline-0:1.5. 100% |  13.0 MiB/s | 146.5 KiB |  00m00s
[163/203] Installing man-db-0:2.13.1-3. 100% |  68.0 MiB/s |   2.9 MiB |  00m00s
[164/203] Installing environment-module 100% |  55.7 MiB/s |   1.9 MiB |  00m00s
[165/203] Installing munge-libs-0:0.5.1 100% |  31.9 MiB/s |  32.7 KiB |  00m00s
[166/203] Installing pmix-0:5.0.7-3.fc4 100% | 281.8 MiB/s |   2.3 MiB |  00m00s
[167/203] Installing prrte-libs-0:3.0.6 100% | 242.5 MiB/s |   1.7 MiB |  00m00s
[168/203] Installing prrte-0:3.0.6-9.fc 100% | 161.5 MiB/s | 165.3 KiB |  00m00s
[169/203] Installing openssh-0:10.2p1-5 100% |  65.1 MiB/s |   1.4 MiB |  00m00s
[170/203] Installing libedit-0:3.1-58.2 100% | 122.0 MiB/s | 249.8 KiB |  00m00s
[171/203] Installing openssh-clients-0: 100% |  85.9 MiB/s |   2.7 MiB |  00m00s
[172/203] Installing git-core-0:2.53.0- 100% | 300.2 MiB/s |  24.6 MiB |  00m00s
[173/203] Installing git-core-doc-0:2.5 100% | 341.1 MiB/s |  18.8 MiB |  00m00s
[174/203] Installing git-0:2.53.0-1.fc4 100% |  56.4 MiB/s |  57.7 KiB |  00m00s
[175/203] Installing perl-Git-0:2.53.0- 100% |  63.8 MiB/s |  65.4 KiB |  00m00s
[176/203] Installing rocm-clang-0:22-5. 100% |  69.1 MiB/s | 147.2 MiB |  00m02s
[177/203] Installing rocm-clang-devel-0 100% | 115.2 MiB/s |  29.6 MiB |  00m00s
[178/203] Installing rocm-device-libs-0 100% |  86.4 MiB/s |   3.3 MiB |  00m00s
[179/203] Installing rocm-comgr-devel-0 100% |  49.1 MiB/s | 100.6 KiB |  00m00s
[180/203] Installing hipcc-0:22-5.rocm7 100% |  22.9 MiB/s | 634.5 KiB |  00m00s
[181/203] Installing rocm-hip-0:7.2.0-3 100% | 340.3 MiB/s |  27.2 MiB |  00m00s
[182/203] Installing rocblas-0:7.2.0-5. 100% |  76.3 MiB/s | 983.4 MiB |  00m13s
[183/203] Installing rocsolver-0:7.2.0- 100% |  26.6 MiB/s |   1.1 GiB |  00m43s
[184/203] Installing hipblas-0:7.2.0-4. 100% |  10.9 MiB/s | 804.4 KiB |  00m00s
[185/203] Installing rocm-hip-devel-0:7 100% |  22.8 MiB/s |   2.5 MiB |  00m00s
[186/203] Installing pthreadpool-0:0.0^ 100% |   8.6 MiB/s | 114.4 KiB |  00m00s
[187/203] Installing rhash-0:1.4.5-4.fc 100% |   8.8 MiB/s | 360.1 KiB |  00m00s
[188/203] Installing libuv-1:1.52.1-1.f 100% |  72.8 MiB/s | 596.0 KiB |  00m00s
[189/203] Installing jsoncpp-0:1.9.6-3. 100% |  25.5 MiB/s | 287.1 KiB |  00m00s
[190/203] Installing cmake-0:4.2.3-2.fc 100% | 151.9 MiB/s |  39.1 MiB |  00m00s
[191/203] Installing cmake-data-0:4.2.3 100% |  98.6 MiB/s |  10.0 MiB |  00m00s
[192/203] Installing pthreadpool-devel- 100% |  32.5 MiB/s |  99.8 KiB |  00m00s
[193/203] Installing rocblas-devel-0:7. 100% | 143.5 MiB/s |   2.7 MiB |  00m00s
[194/203] Installing hipblas-devel-0:7. 100% | 120.2 MiB/s |   2.4 MiB |  00m00s
[195/203] Installing openmpi-0:5.0.10-1 100% | 286.1 MiB/s |   7.2 MiB |  00m00s
[196/203] Installing rocm-rpm-macros-0: 100% |   3.2 MiB/s |  19.5 KiB |  00m00s
[197/203] Installing wget2-wget-0:2.2.1 100% |  20.6 KiB/s | 444.0   B |  00m00s
[198/203] Installing gcc-c++-0:16.0.1-0 100% |  82.7 MiB/s |  46.3 MiB |  00m01s
[199/203] Installing annobin-plugin-gcc 100% |  40.1 MiB/s | 698.8 KiB |  00m00s
[200/203] Installing gcc-plugin-annobin 100% |   2.6 MiB/s |  58.8 KiB |  00m00s
[201/203] Installing libcurl-devel-0:8. 100% |  37.5 MiB/s |   1.5 MiB |  00m00s
[202/203] Installing langpacks-en-0:4.3 100% | 227.9 KiB/s | 700.0   B |  00m00s
[203/203] Installing xxd-2:9.2.148-1.fc 100% |  61.5 KiB/s |  34.2 KiB |  00m01s
Warning: skipped OpenPGP checks for 35 packages from repository: copr_base
Complete!
Building target platforms: x86_64
Building for target x86_64
setting SOURCE_DATE_EPOCH=1773792000
Wrote: /builddir/build/SRPMS/llama-cpp-b8064-1.fc45.src.rpm
Updating and loading repositories:
 fedora                                 100% |  44.6 KiB/s |  21.6 KiB |  00m00s
 Copr repository                        100% |  11.2 KiB/s |   1.5 KiB |  00m00s
Repositories loaded.
Package "cmake-4.2.3-2.fc45.x86_64" is already installed.
Package "curl-8.19.0~rc3-1.fc45.x86_64" is already installed.
Package "gcc-c++-16.0.1-0.9.fc45.x86_64" is already installed.
Package "git-2.53.0-1.fc44.x86_64" is already installed.
Package "hipblas-devel-7.2.0-4.fc45.x86_64" is already installed.
Package "langpacks-en-4.3-1.fc45.noarch" is already installed.
Package "libcurl-devel-8.19.0~rc3-1.fc45.x86_64" is already installed.
Package "openmpi-5.0.10-1.fc45.x86_64" is already installed.
Package "pthreadpool-devel-0.0^git20230829.4fe0e1e-9.fc44.x86_64" is already installed.
Package "rocblas-devel-7.2.0-5.fc45.x86_64" is already installed.
Package "rocm-comgr-devel-22-5.rocm7.2.0.fc45.x86_64" is already installed.
Package "rocm-hip-devel-7.2.0-3.fc45.x86_64" is already installed.
Package "rocm-rpm-macros-7.2.0-1.fc45.noarch" is already installed.
Package "rocm-runtime-devel-7.2.0-3.fc45.x86_64" is already installed.
Package "wget2-wget-2.2.1-2.fc44.x86_64" is already installed.
Package "xxd-2:9.2.148-1.fc45.x86_64" is already installed.

Nothing to do.
Finish: build setup for llama-cpp-b8064-1.fc45.src.rpm
Start: rpmbuild llama-cpp-b8064-1.fc45.src.rpm
Building target platforms: x86_64
Building for target x86_64
setting SOURCE_DATE_EPOCH=1773792000
Executing(%mkbuilddir): /bin/sh -e /var/tmp/rpm-tmp.e4WwPf
Executing(%prep): /bin/sh -e /var/tmp/rpm-tmp.G6aLXD
+ umask 022
+ cd /builddir/build/BUILD/llama-cpp-b8064-build
+ cd /builddir/build/BUILD/llama-cpp-b8064-build
+ rm -rf llama.cpp-b8064
+ /usr/lib/rpm/rpmuncompress -x /builddir/build/SOURCES/llama.cpp-b8064.tar.gz
+ STATUS=0
+ '[' 0 -ne 0 ']'
+ cd llama.cpp-b8064
+ /usr/bin/chmod -Rf a+rX,u+w,g-w,o-w .
+ 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.vIz82I
+ umask 022
+ cd /builddir/build/BUILD/llama-cpp-b8064-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 -mtls-dialect=gnu2 -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 -mtls-dialect=gnu2 -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 -mtls-dialect=gnu2 -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-b8064
+ export HIPCC_COMPILE_FLAGS_APPEND=--offload-compress
+ HIPCC_COMPILE_FLAGS_APPEND=--offload-compress
+ 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 -mtls-dialect=gnu2 -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 -mtls-dialect=gnu2 -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 -mtls-dialect=gnu2 -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 -G Ninja -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 -DINCLUDE_INSTALL_DIR:PATH=/usr/include -DLIB_INSTALL_DIR:PATH=/usr/lib64 -DSYSCONF_INSTALL_DIR:PATH=/etc -DSHARE_INSTALL_PREFIX:PATH=/usr/share -DLIB_SUFFIX=64 -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;gfx950;gfx1010;gfx1012;gfx1030;gfx1031;gfx1035;gfx1036;gfx1100;gfx1101;gfx1102;gfx1103;gfx1150;gfx1151;gfx1152;gfx1153;gfx1200;gfx1201' -DLLAMA_BUILD_EXAMPLES=OFF -DLLAMA_BUILD_TESTS=OFF
-- The C compiler identification is Clang 22.0.0
-- The CXX compiler identification is Clang 22.0.0
-- 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
CMAKE_BUILD_TYPE=Release
-- Found Git: /usr/bin/git (found version "2.53.0")
fatal: not a git repository (or any of the parent directories): .git
fatal: not a git repository (or any of the parent directories): .git
-- The ASM compiler identification is Clang with GNU-like command-line
-- Found assembler: /usr/bin/hipcc
-- 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
-- GGML_SYSTEM_ARCH: x86
-- 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) 
-- x86 detected
-- Adding CPU backend variant ggml-cpu:  
CMake Warning at ggml/src/ggml-cpu/CMakeLists.txt:79 (message):
  OpenMP not found
Call Stack (most recent call first):
  ggml/src/CMakeLists.txt:445 (ggml_add_cpu_backend_variant_impl)


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.


CMake Warning (dev) at /usr/lib64/cmake/hip/hip-config-amd.cmake:70 (message):
  AMDGPU_TARGETS is deprecated.  Please use GPU_TARGETS instead.
Call Stack (most recent call first):
  /usr/lib64/cmake/hip/hip-config.cmake:148 (include)
  ggml/src/ggml-hip/CMakeLists.txt:42 (find_package)
This warning is for project developers.  Use -Wno-dev to suppress it.

-- Performing Test HIP_CLANG_SUPPORTS_PARALLEL_JOBS
-- Performing Test HIP_CLANG_SUPPORTS_PARALLEL_JOBS - Success
-- HIP and hipBLAS found
-- Including HIP backend
-- ggml version: 0.9.5
-- ggml commit:  unknown
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.


-- Found OpenSSL: /usr/lib64/libcrypto.so (found version "3.5.5")
-- Performing Test OPENSSL_VERSION_SUPPORTED
-- Performing Test OPENSSL_VERSION_SUPPORTED - Success
-- OpenSSL found: 3.5.5
-- Generating embedded license file for target: common
-- Configuring done (11.9s)
-- Generating done (0.1s)
CMake Warning:
  Manually-specified variables were not used by the project:

    CMAKE_Fortran_FLAGS_RELEASE
    CMAKE_INSTALL_DO_STRIP
    CMAKE_VERBOSE_MAKEFILE
    INCLUDE_INSTALL_DIR
    LIB_INSTALL_DIR
    LIB_SUFFIX
    SHARE_INSTALL_PREFIX
    SYSCONF_INSTALL_DIR


-- Build files have been written to: /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/redhat-linux-build
+ /usr/bin/cmake --build redhat-linux-build -j4 --verbose
Change Dir: '/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/redhat-linux-build'

Run Build Command(s): /usr/bin/ninja-build -v -j 4
[1/401] /usr/bin/hipcc -DGGML_BUILD -DGGML_COMMIT=\"unknown\" -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_VERSION=\"0.9.5\" -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -mtls-dialect=gnu2 -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 ggml/src/CMakeFiles/ggml-base.dir/ggml-alloc.c.o.d -o ggml/src/CMakeFiles/ggml-base.dir/ggml-alloc.c.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-alloc.c
clang++: warning: argument unused during compilation: '--offload-compress' [-Wunused-command-line-argument]
[2/401] /usr/bin/hipcc -DGGML_BUILD -DGGML_COMMIT=\"unknown\" -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_VERSION=\"0.9.5\" -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -mtls-dialect=gnu2 -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 ggml/src/CMakeFiles/ggml-base.dir/ggml.c.o.d -o ggml/src/CMakeFiles/ggml-base.dir/ggml.c.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml.c
clang++: warning: argument unused during compilation: '--offload-compress' [-Wunused-command-line-argument]
[3/401] /usr/bin/hipcc -DGGML_BUILD -DGGML_COMMIT=\"unknown\" -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_VERSION=\"0.9.5\" -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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.cpp.o -MF ggml/src/CMakeFiles/ggml-base.dir/ggml.cpp.o.d -o ggml/src/CMakeFiles/ggml-base.dir/ggml.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml.cpp
[4/401] /usr/bin/hipcc -DGGML_BUILD -DGGML_COMMIT=\"unknown\" -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_VERSION=\"0.9.5\" -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 ggml/src/CMakeFiles/ggml-base.dir/ggml-backend.cpp.o.d -o ggml/src/CMakeFiles/ggml-base.dir/ggml-backend.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-backend.cpp
[5/401] /usr/bin/hipcc -DGGML_BUILD -DGGML_COMMIT=\"unknown\" -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_VERSION=\"0.9.5\" -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 ggml/src/CMakeFiles/ggml-base.dir/ggml-threading.cpp.o.d -o ggml/src/CMakeFiles/ggml-base.dir/ggml-threading.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-threading.cpp
[6/401] /usr/bin/hipcc -DGGML_BUILD -DGGML_COMMIT=\"unknown\" -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_VERSION=\"0.9.5\" -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 ggml/src/CMakeFiles/ggml-base.dir/ggml-opt.cpp.o.d -o ggml/src/CMakeFiles/ggml-base.dir/ggml-opt.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-opt.cpp
[7/401] /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU_REPACK -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -mtls-dialect=gnu2 -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-cpu.dir/ggml-cpu/ggml-cpu.c.o -MF ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu.c.o.d -o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu.c.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu/ggml-cpu.c
clang++: warning: argument unused during compilation: '--offload-compress' [-Wunused-command-line-argument]
[8/401] /usr/bin/hipcc -DGGML_BUILD -DGGML_COMMIT=\"unknown\" -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_VERSION=\"0.9.5\" -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -mtls-dialect=gnu2 -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 ggml/src/CMakeFiles/ggml-base.dir/ggml-quants.c.o.d -o ggml/src/CMakeFiles/ggml-base.dir/ggml-quants.c.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-quants.c
clang++: warning: argument unused during compilation: '--offload-compress' [-Wunused-command-line-argument]
[9/401] /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU_REPACK -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-cpu.dir/ggml-cpu/hbm.cpp.o -MF ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/hbm.cpp.o.d -o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/hbm.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu/hbm.cpp
[10/401] /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU_REPACK -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-cpu.dir/ggml-cpu/ggml-cpu.cpp.o -MF ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu.cpp.o.d -o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu/ggml-cpu.cpp
[11/401] /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU_REPACK -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -mtls-dialect=gnu2 -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-cpu.dir/ggml-cpu/quants.c.o -MF ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/quants.c.o.d -o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/quants.c.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu/quants.c
clang++: warning: argument unused during compilation: '--offload-compress' [-Wunused-command-line-argument]
[12/401] /usr/bin/hipcc -DGGML_BUILD -DGGML_COMMIT=\"unknown\" -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_VERSION=\"0.9.5\" -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_base_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 ggml/src/CMakeFiles/ggml-base.dir/gguf.cpp.o.d -o ggml/src/CMakeFiles/ggml-base.dir/gguf.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/gguf.cpp
[13/401] /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU_REPACK -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-cpu.dir/ggml-cpu/repack.cpp.o -MF ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/repack.cpp.o.d -o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/repack.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu/repack.cpp
[14/401] /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU_REPACK -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-cpu.dir/ggml-cpu/traits.cpp.o -MF ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/traits.cpp.o.d -o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/traits.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu/traits.cpp
[15/401] /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU_REPACK -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-cpu.dir/ggml-cpu/amx/amx.cpp.o -MF ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/amx/amx.cpp.o.d -o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/amx/amx.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu/amx/amx.cpp
[16/401] /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU_REPACK -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-cpu.dir/ggml-cpu/amx/mmq.cpp.o -MF ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/amx/mmq.cpp.o.d -o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/amx/mmq.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu/amx/mmq.cpp
[17/401] /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU_REPACK -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-cpu.dir/ggml-cpu/binary-ops.cpp.o -MF ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/binary-ops.cpp.o.d -o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/binary-ops.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu/binary-ops.cpp
[18/401] : && /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  -shared -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 -Xlinker --dependency-file=ggml/src/CMakeFiles/ggml-base.dir/link.d -Wl,-soname,libggml-base.so.0 -o bin/libggml-base.so.0.9.5 ggml/src/CMakeFiles/ggml-base.dir/ggml.c.o ggml/src/CMakeFiles/ggml-base.dir/ggml.cpp.o ggml/src/CMakeFiles/ggml-base.dir/ggml-alloc.c.o ggml/src/CMakeFiles/ggml-base.dir/ggml-backend.cpp.o ggml/src/CMakeFiles/ggml-base.dir/ggml-opt.cpp.o ggml/src/CMakeFiles/ggml-base.dir/ggml-threading.cpp.o ggml/src/CMakeFiles/ggml-base.dir/ggml-quants.c.o ggml/src/CMakeFiles/ggml-base.dir/gguf.cpp.o  -lm && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
[19/401] /usr/bin/cmake -E cmake_symlink_library bin/libggml-base.so.0.9.5  bin/libggml-base.so.0 bin/libggml-base.so && :
[20/401] /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU_REPACK -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-cpu.dir/ggml-cpu/vec.cpp.o -MF ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/vec.cpp.o.d -o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/vec.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu/vec.cpp
[21/401] /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU_REPACK -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-cpu.dir/ggml-cpu/unary-ops.cpp.o -MF ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/unary-ops.cpp.o.d -o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/unary-ops.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu/unary-ops.cpp
[22/401] /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU_REPACK -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -mtls-dialect=gnu2 -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-cpu.dir/ggml-cpu/arch/x86/quants.c.o -MF ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/arch/x86/quants.c.o.d -o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/arch/x86/quants.c.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu/arch/x86/quants.c
clang++: warning: argument unused during compilation: '--offload-compress' [-Wunused-command-line-argument]
[23/401] /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU_REPACK -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-cpu.dir/ggml-cpu/llamafile/sgemm.cpp.o -MF ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/llamafile/sgemm.cpp.o.d -o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/llamafile/sgemm.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu/llamafile/sgemm.cpp
[24/401] /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU_REPACK -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-cpu.dir/ggml-cpu/arch/x86/repack.cpp.o -MF ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/arch/x86/repack.cpp.o.d -o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/arch/x86/repack.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu/arch/x86/repack.cpp
[25/401] /usr/bin/hipcc -DGGML_BACKEND_BUILD -DGGML_BACKEND_SHARED -DGGML_SCHED_MAX_COPIES=4 -DGGML_SHARED -DGGML_USE_CPU_REPACK -DGGML_USE_LLAMAFILE -D_GNU_SOURCE -D_XOPEN_SOURCE=600 -Dggml_cpu_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-cpu.dir/ggml-cpu/ops.cpp.o -MF ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ops.cpp.o.d -o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ops.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cpu/ops.cpp
[26/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/acc.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/acc.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/acc.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/acc.cu
[27/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/add-id.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/add-id.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/add-id.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/add-id.cu
[28/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/arange.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/arange.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/arange.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/arange.cu
[29/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/argmax.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/argmax.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/argmax.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/argmax.cu
[30/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/argsort.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/argsort.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/argsort.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/argsort.cu
[31/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/clamp.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/clamp.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/clamp.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/clamp.cu
[32/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/concat.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/concat.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/concat.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
1 warning generated when compiling for gfx1010.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
1 warning generated when compiling for gfx1012.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
1 warning generated when compiling for gfx1030.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
1 warning generated when compiling for gfx1031.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
1 warning generated when compiling for gfx1035.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
1 warning generated when compiling for gfx1036.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
1 warning generated when compiling for gfx1100.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
1 warning generated when compiling for gfx1101.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
1 warning generated when compiling for gfx1102.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
1 warning generated when compiling for gfx1103.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
1 warning generated when compiling for gfx1150.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
1 warning generated when compiling for gfx1151.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
1 warning generated when compiling for gfx1152.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
1 warning generated when compiling for gfx1153.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
1 warning generated when compiling for gfx1200.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
1 warning generated when compiling for gfx1201.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
1 warning generated when compiling for gfx900.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
1 warning generated when compiling for gfx906.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
1 warning generated when compiling for gfx908.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
1 warning generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
1 warning generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
1 warning generated when compiling for gfx942.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
1 warning generated when compiling for gfx950.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/concat.cu:218:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  218 |                 break;
      |                 ^~~~~
1 warning generated when compiling for host.
[33/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv-transpose-1d.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv-transpose-1d.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv-transpose-1d.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv-transpose-1d.cu
[34/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv2d-dw.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv2d-dw.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv2d-dw.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-dw.cu
[35/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv2d-transpose.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv2d-transpose.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv2d-transpose.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:17: warning: no previous prototype for function 'conv2d_transpose_kernel' [-Wmissing-prototypes]
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:12: note: declare 'static' if the function is not intended to be used outside of this translation unit
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |            ^
      |            static 
1 warning generated when compiling for gfx1010.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:17: warning: no previous prototype for function 'conv2d_transpose_kernel' [-Wmissing-prototypes]
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:12: note: declare 'static' if the function is not intended to be used outside of this translation unit
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |            ^
      |            static 
1 warning generated when compiling for gfx1012.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:17: warning: no previous prototype for function 'conv2d_transpose_kernel' [-Wmissing-prototypes]
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:12: note: declare 'static' if the function is not intended to be used outside of this translation unit
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |            ^
      |            static 
1 warning generated when compiling for gfx1030.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:17: warning: no previous prototype for function 'conv2d_transpose_kernel' [-Wmissing-prototypes]
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:12: note: declare 'static' if the function is not intended to be used outside of this translation unit
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |            ^
      |            static 
1 warning generated when compiling for gfx1031.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:17: warning: no previous prototype for function 'conv2d_transpose_kernel' [-Wmissing-prototypes]
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:12: note: declare 'static' if the function is not intended to be used outside of this translation unit
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |            ^
      |            static 
1 warning generated when compiling for gfx1035.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:17: warning: no previous prototype for function 'conv2d_transpose_kernel' [-Wmissing-prototypes]
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:12: note: declare 'static' if the function is not intended to be used outside of this translation unit
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |            ^
      |            static 
1 warning generated when compiling for gfx1036.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:17: warning: no previous prototype for function 'conv2d_transpose_kernel' [-Wmissing-prototypes]
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:12: note: declare 'static' if the function is not intended to be used outside of this translation unit
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |            ^
      |            static 
1 warning generated when compiling for gfx1100.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:17: warning: no previous prototype for function 'conv2d_transpose_kernel' [-Wmissing-prototypes]
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:12: note: declare 'static' if the function is not intended to be used outside of this translation unit
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |            ^
      |            static 
1 warning generated when compiling for gfx1101.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:17: warning: no previous prototype for function 'conv2d_transpose_kernel' [-Wmissing-prototypes]
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:12: note: declare 'static' if the function is not intended to be used outside of this translation unit
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |            ^
      |            static 
1 warning generated when compiling for gfx1102.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:17: warning: no previous prototype for function 'conv2d_transpose_kernel' [-Wmissing-prototypes]
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:12: note: declare 'static' if the function is not intended to be used outside of this translation unit
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |            ^
      |            static 
1 warning generated when compiling for gfx1103.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:17: warning: no previous prototype for function 'conv2d_transpose_kernel' [-Wmissing-prototypes]
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:12: note: declare 'static' if the function is not intended to be used outside of this translation unit
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |            ^
      |            static 
1 warning generated when compiling for gfx1150.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:17: warning: no previous prototype for function 'conv2d_transpose_kernel' [-Wmissing-prototypes]
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:12: note: declare 'static' if the function is not intended to be used outside of this translation unit
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |            ^
      |            static 
1 warning generated when compiling for gfx1151.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:17: warning: no previous prototype for function 'conv2d_transpose_kernel' [-Wmissing-prototypes]
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:12: note: declare 'static' if the function is not intended to be used outside of this translation unit
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |            ^
      |            static 
1 warning generated when compiling for gfx1152.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:17: warning: no previous prototype for function 'conv2d_transpose_kernel' [-Wmissing-prototypes]
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:12: note: declare 'static' if the function is not intended to be used outside of this translation unit
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |            ^
      |            static 
1 warning generated when compiling for gfx1153.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:17: warning: no previous prototype for function 'conv2d_transpose_kernel' [-Wmissing-prototypes]
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:12: note: declare 'static' if the function is not intended to be used outside of this translation unit
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |            ^
      |            static 
1 warning generated when compiling for gfx1200.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:17: warning: no previous prototype for function 'conv2d_transpose_kernel' [-Wmissing-prototypes]
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:12: note: declare 'static' if the function is not intended to be used outside of this translation unit
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |            ^
      |            static 
1 warning generated when compiling for gfx1201.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:17: warning: no previous prototype for function 'conv2d_transpose_kernel' [-Wmissing-prototypes]
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:12: note: declare 'static' if the function is not intended to be used outside of this translation unit
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |            ^
      |            static 
1 warning generated when compiling for gfx900.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:17: warning: no previous prototype for function 'conv2d_transpose_kernel' [-Wmissing-prototypes]
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:12: note: declare 'static' if the function is not intended to be used outside of this translation unit
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |            ^
      |            static 
1 warning generated when compiling for gfx906.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:17: warning: no previous prototype for function 'conv2d_transpose_kernel' [-Wmissing-prototypes]
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:12: note: declare 'static' if the function is not intended to be used outside of this translation unit
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |            ^
      |            static 
1 warning generated when compiling for gfx908.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:17: warning: no previous prototype for function 'conv2d_transpose_kernel' [-Wmissing-prototypes]
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:12: note: declare 'static' if the function is not intended to be used outside of this translation unit
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |            ^
      |            static 
1 warning generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:17: warning: no previous prototype for function 'conv2d_transpose_kernel' [-Wmissing-prototypes]
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:12: note: declare 'static' if the function is not intended to be used outside of this translation unit
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |            ^
      |            static 
1 warning generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:17: warning: no previous prototype for function 'conv2d_transpose_kernel' [-Wmissing-prototypes]
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:12: note: declare 'static' if the function is not intended to be used outside of this translation unit
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |            ^
      |            static 
1 warning generated when compiling for gfx942.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:17: warning: no previous prototype for function 'conv2d_transpose_kernel' [-Wmissing-prototypes]
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:12: note: declare 'static' if the function is not intended to be used outside of this translation unit
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |            ^
      |            static 
1 warning generated when compiling for gfx950.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:17: warning: no previous prototype for function 'conv2d_transpose_kernel' [-Wmissing-prototypes]
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d-transpose.cu:6:12: note: declare 'static' if the function is not intended to be used outside of this translation unit
    6 | __global__ void conv2d_transpose_kernel(const float * __restrict__ input, const half * __restrict__ kernel,
      |            ^
      |            static 
1 warning generated when compiling for host.
[36/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/binbcast.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/binbcast.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/binbcast.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu:502:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  502 |         } break;
      |           ^~~~~
1 warning generated when compiling for gfx1010.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu:502:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  502 |         } break;
      |           ^~~~~
1 warning generated when compiling for gfx1012.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu:502:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  502 |         } break;
      |           ^~~~~
1 warning generated when compiling for gfx1030.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu:502:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  502 |         } break;
      |           ^~~~~
1 warning generated when compiling for gfx1031.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu:502:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  502 |         } break;
      |           ^~~~~
1 warning generated when compiling for gfx1035.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu:502:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  502 |         } break;
      |           ^~~~~
1 warning generated when compiling for gfx1036.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu:502:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  502 |         } break;
      |           ^~~~~
1 warning generated when compiling for gfx1100.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu:502:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  502 |         } break;
      |           ^~~~~
1 warning generated when compiling for gfx1101.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu:502:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  502 |         } break;
      |           ^~~~~
1 warning generated when compiling for gfx1102.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu:502:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  502 |         } break;
      |           ^~~~~
1 warning generated when compiling for gfx1103.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu:502:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  502 |         } break;
      |           ^~~~~
1 warning generated when compiling for gfx1150.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu:502:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  502 |         } break;
      |           ^~~~~
1 warning generated when compiling for gfx1151.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu:502:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  502 |         } break;
      |           ^~~~~
1 warning generated when compiling for gfx1152.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu:502:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  502 |         } break;
      |           ^~~~~
1 warning generated when compiling for gfx1153.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu:502:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  502 |         } break;
      |           ^~~~~
1 warning generated when compiling for gfx1200.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu:502:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  502 |         } break;
      |           ^~~~~
1 warning generated when compiling for gfx1201.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu:502:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  502 |         } break;
      |           ^~~~~
1 warning generated when compiling for gfx900.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu:502:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  502 |         } break;
      |           ^~~~~
1 warning generated when compiling for gfx906.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu:502:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  502 |         } break;
      |           ^~~~~
1 warning generated when compiling for gfx908.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu:502:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  502 |         } break;
      |           ^~~~~
1 warning generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu:502:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  502 |         } break;
      |           ^~~~~
1 warning generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu:502:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  502 |         } break;
      |           ^~~~~
1 warning generated when compiling for gfx942.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu:502:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  502 |         } break;
      |           ^~~~~
1 warning generated when compiling for gfx950.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/binbcast.cu:502:11: warning: 'break' will never be executed [-Wunreachable-code-break]
  502 |         } break;
      |           ^~~~~
1 warning generated when compiling for host.
[37/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/count-equal.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/count-equal.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/count-equal.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1010.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1012.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1030.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1031.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1035.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1036.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1100.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1101.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1102.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1103.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1150.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1151.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1152.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1153.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1200.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1201.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx900.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx906.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx908.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx942.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx950.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/count-equal.cu:62:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   62 |             break;
      |             ^~~~~
1 warning generated when compiling for host.
[38/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv2d.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv2d.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv2d.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/conv2d.cu
[39/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/convert.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/convert.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/convert.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:509:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  509 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:512:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  512 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
3 warnings generated when compiling for gfx1010.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:509:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  509 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:512:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  512 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
3 warnings generated when compiling for gfx1012.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:509:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  509 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:512:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  512 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
3 warnings generated when compiling for gfx1030.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:509:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  509 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:512:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  512 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
3 warnings generated when compiling for gfx1031.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:509:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  509 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:512:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  512 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
3 warnings generated when compiling for gfx1035.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:509:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  509 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:512:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  512 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
3 warnings generated when compiling for gfx1036.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:509:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  509 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:512:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  512 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
3 warnings generated when compiling for gfx1100.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:509:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  509 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:512:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  512 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
3 warnings generated when compiling for gfx1101.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:509:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  509 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:512:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  512 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
3 warnings generated when compiling for gfx1102.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:509:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  509 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:512:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  512 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
3 warnings generated when compiling for gfx1103.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:509:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  509 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:512:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  512 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
3 warnings generated when compiling for gfx1150.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:509:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  509 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:512:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  512 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
3 warnings generated when compiling for gfx1151.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:509:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  509 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:512:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  512 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
3 warnings generated when compiling for gfx1152.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:509:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  509 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:512:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  512 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
3 warnings generated when compiling for gfx1153.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:509:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  509 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:512:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  512 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
3 warnings generated when compiling for gfx1200.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:509:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  509 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:512:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  512 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
3 warnings generated when compiling for gfx1201.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:509:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  509 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:512:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  512 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
3 warnings generated when compiling for gfx900.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:509:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  509 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:512:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  512 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
3 warnings generated when compiling for gfx906.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:509:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  509 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:512:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  512 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
3 warnings generated when compiling for gfx908.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:509:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  509 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:512:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  512 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
3 warnings generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:509:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  509 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:512:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  512 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
3 warnings generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:509:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  509 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:512:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  512 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
3 warnings generated when compiling for gfx942.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:509:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  509 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:512:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  512 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
3 warnings generated when compiling for gfx950.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:509:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<false>' requested here
  509 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:49:31: warning: cast from 'const void *' to 'int *' drops const qualifier [-Wcast-qual]
   49 |     const int * x0 = ((int *) vx) + blockIdx.x * nint;
      |                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/convert.cu:512:9: note: in instantiation of function template specialization 'dequantize_block_q8_0_f16<true>' requested here
  512 |         dequantize_block_q8_0_f16<need_check><<<num_blocks, WARP_SIZE, 0, stream>>>(vx, y, k);
      |         ^
3 warnings generated when compiling for host.
[40/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/cross-entropy-loss.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/cross-entropy-loss.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/cross-entropy-loss.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cross-entropy-loss.cu
[41/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/cumsum.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/cumsum.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/cumsum.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:13:32: warning: unused parameter 'src' [-Wunused-parameter]
   13 |         const T * __restrict__ src,
      |                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:14:26: warning: unused parameter 'dst' [-Wunused-parameter]
   14 |         T * __restrict__ dst,
      |                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:23: warning: unused parameter 'ne00' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:43: warning: unused parameter 'ne01' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:63: warning: unused parameter 'ne02' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:83: warning: unused parameter 'ne03' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:24: warning: unused parameter 's01' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:44: warning: unused parameter 's02' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:64: warning: unused parameter 's03' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:25: warning: unused parameter 's1' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:46: warning: unused parameter 's2' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:67: warning: unused parameter 's3' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                                                   ^
12 warnings generated when compiling for gfx1010.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:13:32: warning: unused parameter 'src' [-Wunused-parameter]
   13 |         const T * __restrict__ src,
      |                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:14:26: warning: unused parameter 'dst' [-Wunused-parameter]
   14 |         T * __restrict__ dst,
      |                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:23: warning: unused parameter 'ne00' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:43: warning: unused parameter 'ne01' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:63: warning: unused parameter 'ne02' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:83: warning: unused parameter 'ne03' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:24: warning: unused parameter 's01' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:44: warning: unused parameter 's02' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:64: warning: unused parameter 's03' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:25: warning: unused parameter 's1' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:46: warning: unused parameter 's2' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:67: warning: unused parameter 's3' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                                                   ^
12 warnings generated when compiling for gfx1012.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:13:32: warning: unused parameter 'src' [-Wunused-parameter]
   13 |         const T * __restrict__ src,
      |                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:14:26: warning: unused parameter 'dst' [-Wunused-parameter]
   14 |         T * __restrict__ dst,
      |                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:23: warning: unused parameter 'ne00' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:43: warning: unused parameter 'ne01' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:63: warning: unused parameter 'ne02' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:83: warning: unused parameter 'ne03' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:24: warning: unused parameter 's01' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:44: warning: unused parameter 's02' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:64: warning: unused parameter 's03' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:25: warning: unused parameter 's1' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:46: warning: unused parameter 's2' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:67: warning: unused parameter 's3' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                                                   ^
12 warnings generated when compiling for gfx1030.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:13:32: warning: unused parameter 'src' [-Wunused-parameter]
   13 |         const T * __restrict__ src,
      |                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:14:26: warning: unused parameter 'dst' [-Wunused-parameter]
   14 |         T * __restrict__ dst,
      |                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:23: warning: unused parameter 'ne00' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:43: warning: unused parameter 'ne01' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:63: warning: unused parameter 'ne02' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:83: warning: unused parameter 'ne03' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:24: warning: unused parameter 's01' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:44: warning: unused parameter 's02' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:64: warning: unused parameter 's03' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:25: warning: unused parameter 's1' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:46: warning: unused parameter 's2' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:67: warning: unused parameter 's3' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                                                   ^
12 warnings generated when compiling for gfx1031.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:13:32: warning: unused parameter 'src' [-Wunused-parameter]
   13 |         const T * __restrict__ src,
      |                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:14:26: warning: unused parameter 'dst' [-Wunused-parameter]
   14 |         T * __restrict__ dst,
      |                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:23: warning: unused parameter 'ne00' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:43: warning: unused parameter 'ne01' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:63: warning: unused parameter 'ne02' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:83: warning: unused parameter 'ne03' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:24: warning: unused parameter 's01' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:44: warning: unused parameter 's02' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:64: warning: unused parameter 's03' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:25: warning: unused parameter 's1' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:46: warning: unused parameter 's2' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:67: warning: unused parameter 's3' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                                                   ^
12 warnings generated when compiling for gfx1035.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:13:32: warning: unused parameter 'src' [-Wunused-parameter]
   13 |         const T * __restrict__ src,
      |                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:14:26: warning: unused parameter 'dst' [-Wunused-parameter]
   14 |         T * __restrict__ dst,
      |                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:23: warning: unused parameter 'ne00' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:43: warning: unused parameter 'ne01' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:63: warning: unused parameter 'ne02' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:83: warning: unused parameter 'ne03' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:24: warning: unused parameter 's01' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:44: warning: unused parameter 's02' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:64: warning: unused parameter 's03' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:25: warning: unused parameter 's1' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:46: warning: unused parameter 's2' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:67: warning: unused parameter 's3' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                                                   ^
12 warnings generated when compiling for gfx1036.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:13:32: warning: unused parameter 'src' [-Wunused-parameter]
   13 |         const T * __restrict__ src,
      |                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:14:26: warning: unused parameter 'dst' [-Wunused-parameter]
   14 |         T * __restrict__ dst,
      |                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:23: warning: unused parameter 'ne00' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:43: warning: unused parameter 'ne01' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:63: warning: unused parameter 'ne02' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:83: warning: unused parameter 'ne03' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:24: warning: unused parameter 's01' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:44: warning: unused parameter 's02' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:64: warning: unused parameter 's03' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:25: warning: unused parameter 's1' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:46: warning: unused parameter 's2' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:67: warning: unused parameter 's3' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                                                   ^
12 warnings generated when compiling for gfx1100.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:13:32: warning: unused parameter 'src' [-Wunused-parameter]
   13 |         const T * __restrict__ src,
      |                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:14:26: warning: unused parameter 'dst' [-Wunused-parameter]
   14 |         T * __restrict__ dst,
      |                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:23: warning: unused parameter 'ne00' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:43: warning: unused parameter 'ne01' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:63: warning: unused parameter 'ne02' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:83: warning: unused parameter 'ne03' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:24: warning: unused parameter 's01' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:44: warning: unused parameter 's02' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:64: warning: unused parameter 's03' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:25: warning: unused parameter 's1' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:46: warning: unused parameter 's2' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:67: warning: unused parameter 's3' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                                                   ^
12 warnings generated when compiling for gfx1101.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:13:32: warning: unused parameter 'src' [-Wunused-parameter]
   13 |         const T * __restrict__ src,
      |                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:14:26: warning: unused parameter 'dst' [-Wunused-parameter]
   14 |         T * __restrict__ dst,
      |                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:23: warning: unused parameter 'ne00' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:43: warning: unused parameter 'ne01' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:63: warning: unused parameter 'ne02' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:83: warning: unused parameter 'ne03' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:24: warning: unused parameter 's01' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:44: warning: unused parameter 's02' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:64: warning: unused parameter 's03' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:25: warning: unused parameter 's1' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:46: warning: unused parameter 's2' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:67: warning: unused parameter 's3' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                                                   ^
12 warnings generated when compiling for gfx1102.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:13:32: warning: unused parameter 'src' [-Wunused-parameter]
   13 |         const T * __restrict__ src,
      |                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:14:26: warning: unused parameter 'dst' [-Wunused-parameter]
   14 |         T * __restrict__ dst,
      |                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:23: warning: unused parameter 'ne00' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:43: warning: unused parameter 'ne01' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:63: warning: unused parameter 'ne02' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:83: warning: unused parameter 'ne03' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:24: warning: unused parameter 's01' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:44: warning: unused parameter 's02' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:64: warning: unused parameter 's03' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:25: warning: unused parameter 's1' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:46: warning: unused parameter 's2' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:67: warning: unused parameter 's3' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                                                   ^
12 warnings generated when compiling for gfx1103.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:13:32: warning: unused parameter 'src' [-Wunused-parameter]
   13 |         const T * __restrict__ src,
      |                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:14:26: warning: unused parameter 'dst' [-Wunused-parameter]
   14 |         T * __restrict__ dst,
      |                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:23: warning: unused parameter 'ne00' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:43: warning: unused parameter 'ne01' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:63: warning: unused parameter 'ne02' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:83: warning: unused parameter 'ne03' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:24: warning: unused parameter 's01' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:44: warning: unused parameter 's02' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:64: warning: unused parameter 's03' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:25: warning: unused parameter 's1' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:46: warning: unused parameter 's2' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:67: warning: unused parameter 's3' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                                                   ^
12 warnings generated when compiling for gfx1150.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:13:32: warning: unused parameter 'src' [-Wunused-parameter]
   13 |         const T * __restrict__ src,
      |                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:14:26: warning: unused parameter 'dst' [-Wunused-parameter]
   14 |         T * __restrict__ dst,
      |                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:23: warning: unused parameter 'ne00' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:43: warning: unused parameter 'ne01' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:63: warning: unused parameter 'ne02' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:83: warning: unused parameter 'ne03' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:24: warning: unused parameter 's01' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:44: warning: unused parameter 's02' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:64: warning: unused parameter 's03' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:25: warning: unused parameter 's1' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:46: warning: unused parameter 's2' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:67: warning: unused parameter 's3' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                                                   ^
12 warnings generated when compiling for gfx1151.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:13:32: warning: unused parameter 'src' [-Wunused-parameter]
   13 |         const T * __restrict__ src,
      |                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:14:26: warning: unused parameter 'dst' [-Wunused-parameter]
   14 |         T * __restrict__ dst,
      |                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:23: warning: unused parameter 'ne00' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:43: warning: unused parameter 'ne01' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:63: warning: unused parameter 'ne02' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:83: warning: unused parameter 'ne03' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:24: warning: unused parameter 's01' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:44: warning: unused parameter 's02' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:64: warning: unused parameter 's03' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:25: warning: unused parameter 's1' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:46: warning: unused parameter 's2' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:67: warning: unused parameter 's3' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                                                   ^
12 warnings generated when compiling for gfx1152.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:13:32: warning: unused parameter 'src' [-Wunused-parameter]
   13 |         const T * __restrict__ src,
      |                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:14:26: warning: unused parameter 'dst' [-Wunused-parameter]
   14 |         T * __restrict__ dst,
      |                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:23: warning: unused parameter 'ne00' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:43: warning: unused parameter 'ne01' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:63: warning: unused parameter 'ne02' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:83: warning: unused parameter 'ne03' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:24: warning: unused parameter 's01' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:44: warning: unused parameter 's02' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:64: warning: unused parameter 's03' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:25: warning: unused parameter 's1' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:46: warning: unused parameter 's2' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:67: warning: unused parameter 's3' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                                                   ^
12 warnings generated when compiling for gfx1153.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:13:32: warning: unused parameter 'src' [-Wunused-parameter]
   13 |         const T * __restrict__ src,
      |                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:14:26: warning: unused parameter 'dst' [-Wunused-parameter]
   14 |         T * __restrict__ dst,
      |                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:23: warning: unused parameter 'ne00' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:43: warning: unused parameter 'ne01' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:63: warning: unused parameter 'ne02' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:83: warning: unused parameter 'ne03' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:24: warning: unused parameter 's01' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:44: warning: unused parameter 's02' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:64: warning: unused parameter 's03' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:25: warning: unused parameter 's1' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:46: warning: unused parameter 's2' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:67: warning: unused parameter 's3' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                                                   ^
12 warnings generated when compiling for gfx1200.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:13:32: warning: unused parameter 'src' [-Wunused-parameter]
   13 |         const T * __restrict__ src,
      |                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:14:26: warning: unused parameter 'dst' [-Wunused-parameter]
   14 |         T * __restrict__ dst,
      |                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:23: warning: unused parameter 'ne00' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:43: warning: unused parameter 'ne01' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:63: warning: unused parameter 'ne02' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:83: warning: unused parameter 'ne03' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:24: warning: unused parameter 's01' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:44: warning: unused parameter 's02' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:64: warning: unused parameter 's03' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:25: warning: unused parameter 's1' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:46: warning: unused parameter 's2' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:67: warning: unused parameter 's3' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                                                   ^
12 warnings generated when compiling for gfx1201.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:13:32: warning: unused parameter 'src' [-Wunused-parameter]
   13 |         const T * __restrict__ src,
      |                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:14:26: warning: unused parameter 'dst' [-Wunused-parameter]
   14 |         T * __restrict__ dst,
      |                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:23: warning: unused parameter 'ne00' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:43: warning: unused parameter 'ne01' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:63: warning: unused parameter 'ne02' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:83: warning: unused parameter 'ne03' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:24: warning: unused parameter 's01' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:44: warning: unused parameter 's02' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:64: warning: unused parameter 's03' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:25: warning: unused parameter 's1' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:46: warning: unused parameter 's2' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:67: warning: unused parameter 's3' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                                                   ^
12 warnings generated when compiling for gfx900.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:13:32: warning: unused parameter 'src' [-Wunused-parameter]
   13 |         const T * __restrict__ src,
      |                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:14:26: warning: unused parameter 'dst' [-Wunused-parameter]
   14 |         T * __restrict__ dst,
      |                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:23: warning: unused parameter 'ne00' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:43: warning: unused parameter 'ne01' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:63: warning: unused parameter 'ne02' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:83: warning: unused parameter 'ne03' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:24: warning: unused parameter 's01' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:44: warning: unused parameter 's02' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:64: warning: unused parameter 's03' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:25: warning: unused parameter 's1' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:46: warning: unused parameter 's2' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:67: warning: unused parameter 's3' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                                                   ^
12 warnings generated when compiling for gfx906.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:13:32: warning: unused parameter 'src' [-Wunused-parameter]
   13 |         const T * __restrict__ src,
      |                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:14:26: warning: unused parameter 'dst' [-Wunused-parameter]
   14 |         T * __restrict__ dst,
      |                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:23: warning: unused parameter 'ne00' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:43: warning: unused parameter 'ne01' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:63: warning: unused parameter 'ne02' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:83: warning: unused parameter 'ne03' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:24: warning: unused parameter 's01' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:44: warning: unused parameter 's02' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:64: warning: unused parameter 's03' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:25: warning: unused parameter 's1' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:46: warning: unused parameter 's2' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:67: warning: unused parameter 's3' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                                                   ^
12 warnings generated when compiling for gfx908.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:13:32: warning: unused parameter 'src' [-Wunused-parameter]
   13 |         const T * __restrict__ src,
      |                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:14:26: warning: unused parameter 'dst' [-Wunused-parameter]
   14 |         T * __restrict__ dst,
      |                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:23: warning: unused parameter 'ne00' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:43: warning: unused parameter 'ne01' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:63: warning: unused parameter 'ne02' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:83: warning: unused parameter 'ne03' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:24: warning: unused parameter 's01' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:44: warning: unused parameter 's02' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:64: warning: unused parameter 's03' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:25: warning: unused parameter 's1' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:46: warning: unused parameter 's2' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:67: warning: unused parameter 's3' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                                                   ^
12 warnings generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:13:32: warning: unused parameter 'src' [-Wunused-parameter]
   13 |         const T * __restrict__ src,
      |                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:14:26: warning: unused parameter 'dst' [-Wunused-parameter]
   14 |         T * __restrict__ dst,
      |                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:23: warning: unused parameter 'ne00' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:43: warning: unused parameter 'ne01' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:63: warning: unused parameter 'ne02' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:83: warning: unused parameter 'ne03' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:24: warning: unused parameter 's01' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:44: warning: unused parameter 's02' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:64: warning: unused parameter 's03' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:25: warning: unused parameter 's1' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:46: warning: unused parameter 's2' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:67: warning: unused parameter 's3' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                                                   ^
12 warnings generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:13:32: warning: unused parameter 'src' [-Wunused-parameter]
   13 |         const T * __restrict__ src,
      |                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:14:26: warning: unused parameter 'dst' [-Wunused-parameter]
   14 |         T * __restrict__ dst,
      |                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:23: warning: unused parameter 'ne00' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:43: warning: unused parameter 'ne01' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:63: warning: unused parameter 'ne02' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:83: warning: unused parameter 'ne03' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:24: warning: unused parameter 's01' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:44: warning: unused parameter 's02' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:64: warning: unused parameter 's03' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:25: warning: unused parameter 's1' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:46: warning: unused parameter 's2' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:67: warning: unused parameter 's3' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                                                   ^
12 warnings generated when compiling for gfx942.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:13:32: warning: unused parameter 'src' [-Wunused-parameter]
   13 |         const T * __restrict__ src,
      |                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:14:26: warning: unused parameter 'dst' [-Wunused-parameter]
   14 |         T * __restrict__ dst,
      |                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:23: warning: unused parameter 'ne00' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:43: warning: unused parameter 'ne01' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:63: warning: unused parameter 'ne02' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:83: warning: unused parameter 'ne03' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:24: warning: unused parameter 's01' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:44: warning: unused parameter 's02' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:64: warning: unused parameter 's03' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:25: warning: unused parameter 's1' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:46: warning: unused parameter 's2' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:67: warning: unused parameter 's3' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                                                   ^
12 warnings generated when compiling for gfx950.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:13:32: warning: unused parameter 'src' [-Wunused-parameter]
   13 |         const T * __restrict__ src,
      |                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:14:26: warning: unused parameter 'dst' [-Wunused-parameter]
   14 |         T * __restrict__ dst,
      |                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:23: warning: unused parameter 'ne00' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:43: warning: unused parameter 'ne01' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:63: warning: unused parameter 'ne02' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:15:83: warning: unused parameter 'ne03' [-Wunused-parameter]
   15 |         const int64_t ne00, const int64_t ne01, const int64_t ne02, const int64_t ne03,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:24: warning: unused parameter 's01' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:44: warning: unused parameter 's02' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:16:64: warning: unused parameter 's03' [-Wunused-parameter]
   16 |         const int64_t  s01, const int64_t  s02, const int64_t  s03,
      |                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:25: warning: unused parameter 's1' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:46: warning: unused parameter 's2' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cumsum.cu:17:67: warning: unused parameter 's3' [-Wunused-parameter]
   17 |         const int64_t   s1,  const int64_t   s2,  const int64_t   s3) {
      |                                                                   ^
12 warnings generated when compiling for host.
[42/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/diag.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/diag.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/diag.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/diag.cu
[43/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/diagmask.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/diagmask.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/diagmask.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/diagmask.cu
[44/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/cpy.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/cpy.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/cpy.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cpy.cu
[45/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-tile.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-tile.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-tile.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:47:11: warning: 'break' will never be executed [-Wunreachable-code-break]
   47 |         } break;
      |           ^~~~~
2 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:47:11: warning: 'break' will never be executed [-Wunreachable-code-break]
   47 |         } break;
      |           ^~~~~
2 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:47:11: warning: 'break' will never be executed [-Wunreachable-code-break]
   47 |         } break;
      |           ^~~~~
2 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:47:11: warning: 'break' will never be executed [-Wunreachable-code-break]
   47 |         } break;
      |           ^~~~~
2 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:47:11: warning: 'break' will never be executed [-Wunreachable-code-break]
   47 |         } break;
      |           ^~~~~
2 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:47:11: warning: 'break' will never be executed [-Wunreachable-code-break]
   47 |         } break;
      |           ^~~~~
2 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:47:11: warning: 'break' will never be executed [-Wunreachable-code-break]
   47 |         } break;
      |           ^~~~~
2 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:47:11: warning: 'break' will never be executed [-Wunreachable-code-break]
   47 |         } break;
      |           ^~~~~
2 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:47:11: warning: 'break' will never be executed [-Wunreachable-code-break]
   47 |         } break;
      |           ^~~~~
2 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:47:11: warning: 'break' will never be executed [-Wunreachable-code-break]
   47 |         } break;
      |           ^~~~~
2 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:47:11: warning: 'break' will never be executed [-Wunreachable-code-break]
   47 |         } break;
      |           ^~~~~
2 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:47:11: warning: 'break' will never be executed [-Wunreachable-code-break]
   47 |         } break;
      |           ^~~~~
2 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:47:11: warning: 'break' will never be executed [-Wunreachable-code-break]
   47 |         } break;
      |           ^~~~~
2 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:47:11: warning: 'break' will never be executed [-Wunreachable-code-break]
   47 |         } break;
      |           ^~~~~
2 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:47:11: warning: 'break' will never be executed [-Wunreachable-code-break]
   47 |         } break;
      |           ^~~~~
2 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:47:11: warning: 'break' will never be executed [-Wunreachable-code-break]
   47 |         } break;
      |           ^~~~~
2 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:47:11: warning: 'break' will never be executed [-Wunreachable-code-break]
   47 |         } break;
      |           ^~~~~
2 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:47:11: warning: 'break' will never be executed [-Wunreachable-code-break]
   47 |         } break;
      |           ^~~~~
2 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:47:11: warning: 'break' will never be executed [-Wunreachable-code-break]
   47 |         } break;
      |           ^~~~~
2 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:47:11: warning: 'break' will never be executed [-Wunreachable-code-break]
   47 |         } break;
      |           ^~~~~
2 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:47:11: warning: 'break' will never be executed [-Wunreachable-code-break]
   47 |         } break;
      |           ^~~~~
2 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:47:11: warning: 'break' will never be executed [-Wunreachable-code-break]
   47 |         } break;
      |           ^~~~~
2 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:47:11: warning: 'break' will never be executed [-Wunreachable-code-break]
   47 |         } break;
      |           ^~~~~
2 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cu:47:11: warning: 'break' will never be executed [-Wunreachable-code-break]
   47 |         } break;
      |           ^~~~~
2 warnings generated when compiling for host.
[46/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fill.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fill.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fill.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fill.cu
[47/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:196:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  196 |             break;
      |             ^~~~~
18 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:196:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  196 |             break;
      |             ^~~~~
18 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:196:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  196 |             break;
      |             ^~~~~
18 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:196:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  196 |             break;
      |             ^~~~~
18 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:196:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  196 |             break;
      |             ^~~~~
18 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:196:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  196 |             break;
      |             ^~~~~
18 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:196:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  196 |             break;
      |             ^~~~~
21 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:196:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  196 |             break;
      |             ^~~~~
21 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:196:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  196 |             break;
      |             ^~~~~
21 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:196:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  196 |             break;
      |             ^~~~~
21 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:196:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  196 |             break;
      |             ^~~~~
21 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:196:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  196 |             break;
      |             ^~~~~
21 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:196:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  196 |             break;
      |             ^~~~~
21 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:196:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  196 |             break;
      |             ^~~~~
21 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:196:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  196 |             break;
      |             ^~~~~
26 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:196:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  196 |             break;
      |             ^~~~~
26 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:196:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  196 |             break;
      |             ^~~~~
18 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:196:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  196 |             break;
      |             ^~~~~
18 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:196:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  196 |             break;
      |             ^~~~~
20 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:196:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  196 |             break;
      |             ^~~~~
20 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:196:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  196 |             break;
      |             ^~~~~
20 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:196:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  196 |             break;
      |             ^~~~~
22 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:196:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  196 |             break;
      |             ^~~~~
18 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn.cu:196:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  196 |             break;
      |             ^~~~~
18 warnings generated when compiling for host.
[48/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-wmma-f16.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-wmma-f16.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-wmma-f16.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:7:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:563:15: warning: unused variable 'warp_size' [-Wunused-variable]
  563 |     const int warp_size = ggml_cuda_info().devices[ctx.device].warp_size;
      |               ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:694:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  694 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:667:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  667 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:614:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  614 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:589:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  589 |                     break;
      |                     ^~~~~
6 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:7:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:563:15: warning: unused variable 'warp_size' [-Wunused-variable]
  563 |     const int warp_size = ggml_cuda_info().devices[ctx.device].warp_size;
      |               ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:694:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  694 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:667:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  667 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:614:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  614 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:589:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  589 |                     break;
      |                     ^~~~~
6 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:7:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:563:15: warning: unused variable 'warp_size' [-Wunused-variable]
  563 |     const int warp_size = ggml_cuda_info().devices[ctx.device].warp_size;
      |               ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:694:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  694 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:667:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  667 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:614:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  614 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:589:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  589 |                     break;
      |                     ^~~~~
6 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:7:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:563:15: warning: unused variable 'warp_size' [-Wunused-variable]
  563 |     const int warp_size = ggml_cuda_info().devices[ctx.device].warp_size;
      |               ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:694:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  694 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:667:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  667 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:614:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  614 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:589:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  589 |                     break;
      |                     ^~~~~
6 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:7:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:563:15: warning: unused variable 'warp_size' [-Wunused-variable]
  563 |     const int warp_size = ggml_cuda_info().devices[ctx.device].warp_size;
      |               ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:694:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  694 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:667:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  667 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:614:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  614 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:589:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  589 |                     break;
      |                     ^~~~~
6 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:7:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:563:15: warning: unused variable 'warp_size' [-Wunused-variable]
  563 |     const int warp_size = ggml_cuda_info().devices[ctx.device].warp_size;
      |               ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:694:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  694 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:667:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  667 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:614:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  614 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:589:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  589 |                     break;
      |                     ^~~~~
6 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:7:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:563:15: warning: unused variable 'warp_size' [-Wunused-variable]
  563 |     const int warp_size = ggml_cuda_info().devices[ctx.device].warp_size;
      |               ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:694:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  694 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:667:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  667 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:614:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  614 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:589:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  589 |                     break;
      |                     ^~~~~
6 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:7:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:563:15: warning: unused variable 'warp_size' [-Wunused-variable]
  563 |     const int warp_size = ggml_cuda_info().devices[ctx.device].warp_size;
      |               ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:694:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  694 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:667:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  667 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:614:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  614 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:589:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  589 |                     break;
      |                     ^~~~~
6 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:7:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:563:15: warning: unused variable 'warp_size' [-Wunused-variable]
  563 |     const int warp_size = ggml_cuda_info().devices[ctx.device].warp_size;
      |               ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:694:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  694 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:667:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  667 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:614:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  614 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:589:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  589 |                     break;
      |                     ^~~~~
6 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:7:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:563:15: warning: unused variable 'warp_size' [-Wunused-variable]
  563 |     const int warp_size = ggml_cuda_info().devices[ctx.device].warp_size;
      |               ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:694:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  694 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:667:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  667 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:614:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  614 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:589:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  589 |                     break;
      |                     ^~~~~
6 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:7:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:563:15: warning: unused variable 'warp_size' [-Wunused-variable]
  563 |     const int warp_size = ggml_cuda_info().devices[ctx.device].warp_size;
      |               ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:694:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  694 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:667:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  667 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:614:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  614 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:589:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  589 |                     break;
      |                     ^~~~~
6 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:7:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:563:15: warning: unused variable 'warp_size' [-Wunused-variable]
  563 |     const int warp_size = ggml_cuda_info().devices[ctx.device].warp_size;
      |               ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:694:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  694 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:667:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  667 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:614:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  614 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:589:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  589 |                     break;
      |                     ^~~~~
6 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:7:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:563:15: warning: unused variable 'warp_size' [-Wunused-variable]
  563 |     const int warp_size = ggml_cuda_info().devices[ctx.device].warp_size;
      |               ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:694:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  694 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:667:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  667 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:614:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  614 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:589:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  589 |                     break;
      |                     ^~~~~
6 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:7:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:563:15: warning: unused variable 'warp_size' [-Wunused-variable]
  563 |     const int warp_size = ggml_cuda_info().devices[ctx.device].warp_size;
      |               ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:694:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  694 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:667:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  667 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:614:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  614 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:589:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  589 |                     break;
      |                     ^~~~~
6 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:7:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:563:15: warning: unused variable 'warp_size' [-Wunused-variable]
  563 |     const int warp_size = ggml_cuda_info().devices[ctx.device].warp_size;
      |               ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:694:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  694 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:667:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  667 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:614:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  614 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:589:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  589 |                     break;
      |                     ^~~~~
6 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:7:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:563:15: warning: unused variable 'warp_size' [-Wunused-variable]
  563 |     const int warp_size = ggml_cuda_info().devices[ctx.device].warp_size;
      |               ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:694:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  694 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:667:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  667 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:614:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  614 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:589:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  589 |                     break;
      |                     ^~~~~
6 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:7:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:563:15: warning: unused variable 'warp_size' [-Wunused-variable]
  563 |     const int warp_size = ggml_cuda_info().devices[ctx.device].warp_size;
      |               ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:694:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  694 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:667:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  667 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:614:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  614 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:589:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  589 |                     break;
      |                     ^~~~~
6 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:7:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:563:15: warning: unused variable 'warp_size' [-Wunused-variable]
  563 |     const int warp_size = ggml_cuda_info().devices[ctx.device].warp_size;
      |               ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:694:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  694 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:667:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  667 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:614:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  614 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:589:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  589 |                     break;
      |                     ^~~~~
6 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:7:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:563:15: warning: unused variable 'warp_size' [-Wunused-variable]
  563 |     const int warp_size = ggml_cuda_info().devices[ctx.device].warp_size;
      |               ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:694:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  694 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:667:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  667 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:614:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  614 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:589:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  589 |                     break;
      |                     ^~~~~
6 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:7:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:563:15: warning: unused variable 'warp_size' [-Wunused-variable]
  563 |     const int warp_size = ggml_cuda_info().devices[ctx.device].warp_size;
      |               ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:694:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  694 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:667:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  667 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:614:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  614 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:589:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  589 |                     break;
      |                     ^~~~~
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:7:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:563:15: warning: unused variable 'warp_size' [-Wunused-variable]
  563 |     const int warp_size = ggml_cuda_info().devices[ctx.device].warp_size;
      |               ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:694:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  694 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:667:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  667 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:614:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  614 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:589:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  589 |                     break;
      |                     ^~~~~
6 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:7:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:563:15: warning: unused variable 'warp_size' [-Wunused-variable]
  563 |     const int warp_size = ggml_cuda_info().devices[ctx.device].warp_size;
      |               ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:694:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  694 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:667:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  667 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:614:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  614 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:589:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  589 |                     break;
      |                     ^~~~~
6 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:7:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:563:15: warning: unused variable 'warp_size' [-Wunused-variable]
  563 |     const int warp_size = ggml_cuda_info().devices[ctx.device].warp_size;
      |               ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:694:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  694 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:667:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  667 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:614:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  614 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:589:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  589 |                     break;
      |                     ^~~~~
6 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:7:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:563:15: warning: unused variable 'warp_size' [-Wunused-variable]
  563 |     const int warp_size = ggml_cuda_info().devices[ctx.device].warp_size;
      |               ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:694:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  694 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:667:17: warning: 'break' will never be executed [-Wunreachable-code-break]
  667 |                 break;
      |                 ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:614:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  614 |                     break;
      |                     ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/fattn-wmma-f16.cu:589:21: warning: 'break' will never be executed [-Wunreachable-code-break]
  589 |                     break;
      |                     ^~~~~
6 warnings generated when compiling for host.
[49/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/getrows.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/getrows.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/getrows.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu:234:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  234 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1010.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu:234:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  234 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1012.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu:234:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  234 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1030.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu:234:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  234 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1031.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu:234:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  234 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1035.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu:234:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  234 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1036.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu:234:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  234 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1100.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu:234:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  234 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1101.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu:234:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  234 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1102.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu:234:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  234 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1103.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu:234:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  234 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1150.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu:234:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  234 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1151.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu:234:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  234 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1152.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu:234:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  234 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1153.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu:234:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  234 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1200.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu:234:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  234 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx1201.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu:234:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  234 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx900.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu:234:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  234 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx906.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu:234:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  234 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx908.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu:234:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  234 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu:234:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  234 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu:234:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  234 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx942.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu:234:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  234 |             break;
      |             ^~~~~
1 warning generated when compiling for gfx950.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/getrows.cu:234:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  234 |             break;
      |             ^~~~~
1 warning generated when compiling for host.
[50/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/ggml-cuda.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/ggml-cuda.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/ggml-cuda.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  155 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:154:9: note: declared here
  154 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4212:82: warning: format specifies type 'void *' but the argument has type 'const ggml_tensor *' [-Wformat-pedantic]
 4212 |                 GGML_LOG_DEBUG("Adding stream at node %s %p\n", root_node->name, root_node);
      |                                                          ~~                      ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-impl.h:116:69: note: expanded from macro 'GGML_LOG_DEBUG'
  116 | #define GGML_LOG_DEBUG(...) ggml_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
      |                                                                     ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4768:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4768 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4761:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4761 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4758:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4758 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4749:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4749 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4742:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4742 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4737:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4737 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4732:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4732 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4727:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4727 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4671:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4671 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4664:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4664 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4656:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4656 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4652:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4652 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4633:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4633 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4568:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4568 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4555:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4555 |             break;
      |             ^~~~~
44 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  155 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:154:9: note: declared here
  154 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4212:82: warning: format specifies type 'void *' but the argument has type 'const ggml_tensor *' [-Wformat-pedantic]
 4212 |                 GGML_LOG_DEBUG("Adding stream at node %s %p\n", root_node->name, root_node);
      |                                                          ~~                      ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-impl.h:116:69: note: expanded from macro 'GGML_LOG_DEBUG'
  116 | #define GGML_LOG_DEBUG(...) ggml_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
      |                                                                     ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4768:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4768 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4761:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4761 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4758:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4758 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4749:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4749 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4742:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4742 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4737:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4737 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4732:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4732 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4727:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4727 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4671:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4671 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4664:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4664 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4656:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4656 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4652:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4652 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4633:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4633 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4568:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4568 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4555:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4555 |             break;
      |             ^~~~~
44 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  155 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:154:9: note: declared here
  154 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4212:82: warning: format specifies type 'void *' but the argument has type 'const ggml_tensor *' [-Wformat-pedantic]
 4212 |                 GGML_LOG_DEBUG("Adding stream at node %s %p\n", root_node->name, root_node);
      |                                                          ~~                      ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-impl.h:116:69: note: expanded from macro 'GGML_LOG_DEBUG'
  116 | #define GGML_LOG_DEBUG(...) ggml_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
      |                                                                     ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4768:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4768 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4761:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4761 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4758:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4758 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4749:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4749 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4742:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4742 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4737:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4737 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4732:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4732 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4727:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4727 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4671:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4671 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4664:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4664 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4656:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4656 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4652:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4652 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4633:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4633 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4568:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4568 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4555:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4555 |             break;
      |             ^~~~~
44 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  155 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:154:9: note: declared here
  154 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4212:82: warning: format specifies type 'void *' but the argument has type 'const ggml_tensor *' [-Wformat-pedantic]
 4212 |                 GGML_LOG_DEBUG("Adding stream at node %s %p\n", root_node->name, root_node);
      |                                                          ~~                      ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-impl.h:116:69: note: expanded from macro 'GGML_LOG_DEBUG'
  116 | #define GGML_LOG_DEBUG(...) ggml_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
      |                                                                     ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4768:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4768 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4761:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4761 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4758:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4758 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4749:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4749 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4742:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4742 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4737:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4737 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4732:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4732 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4727:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4727 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4671:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4671 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4664:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4664 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4656:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4656 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4652:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4652 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4633:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4633 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4568:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4568 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4555:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4555 |             break;
      |             ^~~~~
44 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  155 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:154:9: note: declared here
  154 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4212:82: warning: format specifies type 'void *' but the argument has type 'const ggml_tensor *' [-Wformat-pedantic]
 4212 |                 GGML_LOG_DEBUG("Adding stream at node %s %p\n", root_node->name, root_node);
      |                                                          ~~                      ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-impl.h:116:69: note: expanded from macro 'GGML_LOG_DEBUG'
  116 | #define GGML_LOG_DEBUG(...) ggml_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
      |                                                                     ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4768:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4768 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4761:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4761 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4758:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4758 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4749:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4749 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4742:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4742 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4737:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4737 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4732:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4732 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4727:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4727 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4671:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4671 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4664:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4664 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4656:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4656 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4652:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4652 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4633:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4633 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4568:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4568 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4555:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4555 |             break;
      |             ^~~~~
44 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  155 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:154:9: note: declared here
  154 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4212:82: warning: format specifies type 'void *' but the argument has type 'const ggml_tensor *' [-Wformat-pedantic]
 4212 |                 GGML_LOG_DEBUG("Adding stream at node %s %p\n", root_node->name, root_node);
      |                                                          ~~                      ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-impl.h:116:69: note: expanded from macro 'GGML_LOG_DEBUG'
  116 | #define GGML_LOG_DEBUG(...) ggml_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
      |                                                                     ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4768:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4768 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4761:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4761 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4758:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4758 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4749:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4749 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4742:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4742 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4737:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4737 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4732:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4732 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4727:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4727 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4671:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4671 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4664:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4664 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4656:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4656 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4652:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4652 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4633:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4633 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4568:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4568 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4555:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4555 |             break;
      |             ^~~~~
44 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:29:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  155 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:154:9: note: declared here
  154 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4212:82: warning: format specifies type 'void *' but the argument has type 'const ggml_tensor *' [-Wformat-pedantic]
 4212 |                 GGML_LOG_DEBUG("Adding stream at node %s %p\n", root_node->name, root_node);
      |                                                          ~~                      ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-impl.h:116:69: note: expanded from macro 'GGML_LOG_DEBUG'
  116 | #define GGML_LOG_DEBUG(...) ggml_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
      |                                                                     ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4768:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4768 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4761:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4761 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4758:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4758 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4749:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4749 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4742:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4742 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4737:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4737 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4732:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4732 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4727:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4727 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4671:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4671 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4664:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4664 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4656:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4656 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4652:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4652 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4633:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4633 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4568:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4568 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4555:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4555 |             break;
      |             ^~~~~
48 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:29:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  155 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:154:9: note: declared here
  154 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4212:82: warning: format specifies type 'void *' but the argument has type 'const ggml_tensor *' [-Wformat-pedantic]
 4212 |                 GGML_LOG_DEBUG("Adding stream at node %s %p\n", root_node->name, root_node);
      |                                                          ~~                      ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-impl.h:116:69: note: expanded from macro 'GGML_LOG_DEBUG'
  116 | #define GGML_LOG_DEBUG(...) ggml_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
      |                                                                     ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4768:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4768 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4761:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4761 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4758:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4758 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4749:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4749 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4742:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4742 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4737:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4737 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4732:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4732 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4727:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4727 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4671:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4671 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4664:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4664 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4656:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4656 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4652:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4652 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4633:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4633 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4568:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4568 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4555:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4555 |             break;
      |             ^~~~~
48 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:29:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  155 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:154:9: note: declared here
  154 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4212:82: warning: format specifies type 'void *' but the argument has type 'const ggml_tensor *' [-Wformat-pedantic]
 4212 |                 GGML_LOG_DEBUG("Adding stream at node %s %p\n", root_node->name, root_node);
      |                                                          ~~                      ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-impl.h:116:69: note: expanded from macro 'GGML_LOG_DEBUG'
  116 | #define GGML_LOG_DEBUG(...) ggml_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
      |                                                                     ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4768:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4768 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4761:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4761 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4758:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4758 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4749:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4749 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4742:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4742 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4737:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4737 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4732:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4732 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4727:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4727 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4671:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4671 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4664:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4664 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4656:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4656 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4652:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4652 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4633:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4633 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4568:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4568 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4555:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4555 |             break;
      |             ^~~~~
48 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:29:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  155 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:154:9: note: declared here
  154 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4212:82: warning: format specifies type 'void *' but the argument has type 'const ggml_tensor *' [-Wformat-pedantic]
 4212 |                 GGML_LOG_DEBUG("Adding stream at node %s %p\n", root_node->name, root_node);
      |                                                          ~~                      ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-impl.h:116:69: note: expanded from macro 'GGML_LOG_DEBUG'
  116 | #define GGML_LOG_DEBUG(...) ggml_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
      |                                                                     ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4768:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4768 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4761:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4761 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4758:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4758 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4749:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4749 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4742:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4742 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4737:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4737 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4732:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4732 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4727:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4727 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4671:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4671 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4664:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4664 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4656:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4656 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4652:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4652 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4633:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4633 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4568:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4568 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4555:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4555 |             break;
      |             ^~~~~
48 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:29:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  155 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:154:9: note: declared here
  154 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4212:82: warning: format specifies type 'void *' but the argument has type 'const ggml_tensor *' [-Wformat-pedantic]
 4212 |                 GGML_LOG_DEBUG("Adding stream at node %s %p\n", root_node->name, root_node);
      |                                                          ~~                      ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-impl.h:116:69: note: expanded from macro 'GGML_LOG_DEBUG'
  116 | #define GGML_LOG_DEBUG(...) ggml_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
      |                                                                     ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4768:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4768 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4761:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4761 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4758:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4758 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4749:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4749 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4742:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4742 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4737:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4737 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4732:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4732 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4727:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4727 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4671:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4671 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4664:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4664 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4656:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4656 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4652:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4652 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4633:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4633 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4568:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4568 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4555:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4555 |             break;
      |             ^~~~~
48 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:29:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  155 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:154:9: note: declared here
  154 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4212:82: warning: format specifies type 'void *' but the argument has type 'const ggml_tensor *' [-Wformat-pedantic]
 4212 |                 GGML_LOG_DEBUG("Adding stream at node %s %p\n", root_node->name, root_node);
      |                                                          ~~                      ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-impl.h:116:69: note: expanded from macro 'GGML_LOG_DEBUG'
  116 | #define GGML_LOG_DEBUG(...) ggml_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
      |                                                                     ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4768:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4768 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4761:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4761 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4758:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4758 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4749:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4749 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4742:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4742 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4737:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4737 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4732:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4732 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4727:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4727 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4671:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4671 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4664:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4664 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4656:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4656 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4652:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4652 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4633:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4633 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4568:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4568 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4555:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4555 |             break;
      |             ^~~~~
48 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:29:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  155 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:154:9: note: declared here
  154 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4212:82: warning: format specifies type 'void *' but the argument has type 'const ggml_tensor *' [-Wformat-pedantic]
 4212 |                 GGML_LOG_DEBUG("Adding stream at node %s %p\n", root_node->name, root_node);
      |                                                          ~~                      ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-impl.h:116:69: note: expanded from macro 'GGML_LOG_DEBUG'
  116 | #define GGML_LOG_DEBUG(...) ggml_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
      |                                                                     ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4768:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4768 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4761:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4761 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4758:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4758 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4749:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4749 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4742:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4742 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4737:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4737 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4732:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4732 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4727:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4727 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4671:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4671 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4664:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4664 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4656:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4656 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4652:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4652 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4633:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4633 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4568:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4568 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4555:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4555 |             break;
      |             ^~~~~
48 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:29:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  155 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:154:9: note: declared here
  154 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4212:82: warning: format specifies type 'void *' but the argument has type 'const ggml_tensor *' [-Wformat-pedantic]
 4212 |                 GGML_LOG_DEBUG("Adding stream at node %s %p\n", root_node->name, root_node);
      |                                                          ~~                      ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-impl.h:116:69: note: expanded from macro 'GGML_LOG_DEBUG'
  116 | #define GGML_LOG_DEBUG(...) ggml_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
      |                                                                     ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4768:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4768 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4761:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4761 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4758:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4758 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4749:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4749 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4742:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4742 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4737:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4737 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4732:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4732 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4727:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4727 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4671:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4671 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4664:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4664 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4656:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4656 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4652:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4652 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4633:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4633 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4568:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4568 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4555:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4555 |             break;
      |             ^~~~~
48 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:29:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  155 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:154:9: note: declared here
  154 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4212:82: warning: format specifies type 'void *' but the argument has type 'const ggml_tensor *' [-Wformat-pedantic]
 4212 |                 GGML_LOG_DEBUG("Adding stream at node %s %p\n", root_node->name, root_node);
      |                                                          ~~                      ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-impl.h:116:69: note: expanded from macro 'GGML_LOG_DEBUG'
  116 | #define GGML_LOG_DEBUG(...) ggml_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
      |                                                                     ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4768:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4768 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4761:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4761 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4758:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4758 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4749:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4749 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4742:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4742 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4737:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4737 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4732:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4732 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4727:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4727 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4671:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4671 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4664:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4664 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4656:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4656 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4652:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4652 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4633:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4633 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4568:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4568 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4555:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4555 |             break;
      |             ^~~~~
46 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:29:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  155 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:154:9: note: declared here
  154 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4212:82: warning: format specifies type 'void *' but the argument has type 'const ggml_tensor *' [-Wformat-pedantic]
 4212 |                 GGML_LOG_DEBUG("Adding stream at node %s %p\n", root_node->name, root_node);
      |                                                          ~~                      ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-impl.h:116:69: note: expanded from macro 'GGML_LOG_DEBUG'
  116 | #define GGML_LOG_DEBUG(...) ggml_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
      |                                                                     ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4768:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4768 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4761:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4761 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4758:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4758 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4749:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4749 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4742:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4742 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4737:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4737 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4732:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4732 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4727:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4727 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4671:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4671 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4664:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4664 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4656:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4656 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4652:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4652 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4633:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4633 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4568:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4568 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4555:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4555 |             break;
      |             ^~~~~
46 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  155 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:154:9: note: declared here
  154 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4212:82: warning: format specifies type 'void *' but the argument has type 'const ggml_tensor *' [-Wformat-pedantic]
 4212 |                 GGML_LOG_DEBUG("Adding stream at node %s %p\n", root_node->name, root_node);
      |                                                          ~~                      ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-impl.h:116:69: note: expanded from macro 'GGML_LOG_DEBUG'
  116 | #define GGML_LOG_DEBUG(...) ggml_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
      |                                                                     ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4768:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4768 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4761:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4761 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4758:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4758 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4749:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4749 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4742:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4742 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4737:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4737 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4732:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4732 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4727:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4727 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4671:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4671 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4664:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4664 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4656:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4656 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4652:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4652 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4633:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4633 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4568:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4568 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4555:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4555 |             break;
      |             ^~~~~
44 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  155 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:154:9: note: declared here
  154 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4212:82: warning: format specifies type 'void *' but the argument has type 'const ggml_tensor *' [-Wformat-pedantic]
 4212 |                 GGML_LOG_DEBUG("Adding stream at node %s %p\n", root_node->name, root_node);
      |                                                          ~~                      ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-impl.h:116:69: note: expanded from macro 'GGML_LOG_DEBUG'
  116 | #define GGML_LOG_DEBUG(...) ggml_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
      |                                                                     ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4768:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4768 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4761:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4761 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4758:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4758 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4749:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4749 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4742:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4742 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4737:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4737 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4732:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4732 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4727:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4727 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4671:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4671 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4664:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4664 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4656:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4656 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4652:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4652 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4633:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4633 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4568:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4568 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4555:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4555 |             break;
      |             ^~~~~
44 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:29:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  155 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:154:9: note: declared here
  154 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4212:82: warning: format specifies type 'void *' but the argument has type 'const ggml_tensor *' [-Wformat-pedantic]
 4212 |                 GGML_LOG_DEBUG("Adding stream at node %s %p\n", root_node->name, root_node);
      |                                                          ~~                      ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-impl.h:116:69: note: expanded from macro 'GGML_LOG_DEBUG'
  116 | #define GGML_LOG_DEBUG(...) ggml_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
      |                                                                     ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4768:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4768 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4761:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4761 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4758:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4758 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4749:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4749 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4742:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4742 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4737:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4737 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4732:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4732 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4727:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4727 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4671:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4671 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4664:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4664 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4656:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4656 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4652:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4652 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4633:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4633 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4568:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4568 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4555:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4555 |             break;
      |             ^~~~~
47 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:29:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  155 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:154:9: note: declared here
  154 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4212:82: warning: format specifies type 'void *' but the argument has type 'const ggml_tensor *' [-Wformat-pedantic]
 4212 |                 GGML_LOG_DEBUG("Adding stream at node %s %p\n", root_node->name, root_node);
      |                                                          ~~                      ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-impl.h:116:69: note: expanded from macro 'GGML_LOG_DEBUG'
  116 | #define GGML_LOG_DEBUG(...) ggml_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
      |                                                                     ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4768:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4768 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4761:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4761 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4758:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4758 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4749:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4749 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4742:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4742 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4737:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4737 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4732:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4732 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4727:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4727 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4671:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4671 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4664:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4664 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4656:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4656 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4652:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4652 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4633:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4633 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4568:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4568 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4555:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4555 |             break;
      |             ^~~~~
47 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:29:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  155 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:154:9: note: declared here
  154 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4212:82: warning: format specifies type 'void *' but the argument has type 'const ggml_tensor *' [-Wformat-pedantic]
 4212 |                 GGML_LOG_DEBUG("Adding stream at node %s %p\n", root_node->name, root_node);
      |                                                          ~~                      ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-impl.h:116:69: note: expanded from macro 'GGML_LOG_DEBUG'
  116 | #define GGML_LOG_DEBUG(...) ggml_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
      |                                                                     ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4768:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4768 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4761:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4761 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4758:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4758 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4749:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4749 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4742:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4742 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4737:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4737 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4732:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4732 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4727:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4727 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4671:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4671 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4664:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4664 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4656:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4656 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4652:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4652 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4633:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4633 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4568:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4568 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4555:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4555 |             break;
      |             ^~~~~
47 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:29:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  155 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:154:9: note: declared here
  154 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4212:82: warning: format specifies type 'void *' but the argument has type 'const ggml_tensor *' [-Wformat-pedantic]
 4212 |                 GGML_LOG_DEBUG("Adding stream at node %s %p\n", root_node->name, root_node);
      |                                                          ~~                      ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-impl.h:116:69: note: expanded from macro 'GGML_LOG_DEBUG'
  116 | #define GGML_LOG_DEBUG(...) ggml_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
      |                                                                     ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4768:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4768 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4761:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4761 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4758:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4758 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4749:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4749 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4742:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4742 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4737:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4737 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4732:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4732 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4727:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4727 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4671:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4671 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4664:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4664 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4656:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4656 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4652:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4652 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4633:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4633 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4568:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4568 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4555:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4555 |             break;
      |             ^~~~~
49 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  155 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:154:9: note: declared here
  154 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4212:82: warning: format specifies type 'void *' but the argument has type 'const ggml_tensor *' [-Wformat-pedantic]
 4212 |                 GGML_LOG_DEBUG("Adding stream at node %s %p\n", root_node->name, root_node);
      |                                                          ~~                      ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-impl.h:116:69: note: expanded from macro 'GGML_LOG_DEBUG'
  116 | #define GGML_LOG_DEBUG(...) ggml_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
      |                                                                     ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4768:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4768 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4761:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4761 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4758:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4758 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4749:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4749 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4742:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4742 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4737:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4737 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4732:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4732 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4727:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4727 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4671:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4671 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4664:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4664 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4656:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4656 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4652:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4652 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4633:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4633 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4568:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4568 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4555:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4555 |             break;
      |             ^~~~~
44 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:28:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: warning: variable length arrays in C++ are a Clang extension [-Wvla-cxx-extension]
  155 |     char archName[archLen + 1];
      |                   ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:155:19: note: read of non-const variable 'archLen' is not allowed in a constant expression
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:154:9: note: declared here
  154 |     int archLen = strlen(devName);
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4212:82: warning: format specifies type 'void *' but the argument has type 'const ggml_tensor *' [-Wformat-pedantic]
 4212 |                 GGML_LOG_DEBUG("Adding stream at node %s %p\n", root_node->name, root_node);
      |                                                          ~~                      ^~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/../ggml-impl.h:116:69: note: expanded from macro 'GGML_LOG_DEBUG'
  116 | #define GGML_LOG_DEBUG(...) ggml_log_internal(GGML_LOG_LEVEL_DEBUG, __VA_ARGS__)
      |                                                                     ^~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4768:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4768 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4761:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4761 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4758:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4758 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4749:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4749 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4742:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4742 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4737:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4737 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4732:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4732 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4727:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4727 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4671:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4671 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4664:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4664 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4656:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4656 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4652:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4652 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4633:15: warning: 'break' will never be executed [-Wunreachable-code-break]
 4633 |             } break;
      |               ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4568:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4568 |             break;
      |             ^~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ggml-cuda.cu:4555:13: warning: 'break' will never be executed [-Wunreachable-code-break]
 4555 |             break;
      |             ^~~~~
44 warnings generated when compiling for host.
[51/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/gla.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/gla.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/gla.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/gla.cu
[52/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/im2col.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/im2col.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/im2col.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/im2col.cu
[53/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mean.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mean.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mean.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mean.cu
[54/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmf.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmf.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmf.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
27 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
27 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
27 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
27 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
27 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
27 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
30 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
30 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
30 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
30 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
30 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
30 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
30 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
30 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
28 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
28 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
27 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
27 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
29 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
29 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
29 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
31 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
27 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
27 warnings generated when compiling for host.
[55/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmid.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmid.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmid.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmid.cu
[56/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmq.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmq.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmq.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:67:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:67:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:67:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:67:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:67:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:67:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:67:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |             break;
      |             ^~~~~
16 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:67:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |             break;
      |             ^~~~~
16 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:67:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |             break;
      |             ^~~~~
16 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:67:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |             break;
      |             ^~~~~
16 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:67:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |             break;
      |             ^~~~~
16 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:67:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |             break;
      |             ^~~~~
16 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:67:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |             break;
      |             ^~~~~
16 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:67:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |             break;
      |             ^~~~~
16 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:67:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |             break;
      |             ^~~~~
14 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:67:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |             break;
      |             ^~~~~
14 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:67:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:67:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:67:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:67:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:67:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:67:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |             break;
      |             ^~~~~
17 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:67:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cu:67:13: warning: 'break' will never be executed [-Wunreachable-code-break]
   67 |             break;
      |             ^~~~~
12 warnings generated when compiling for host.
[57/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/opt-step-adamw.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/opt-step-adamw.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/opt-step-adamw.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/opt-step-adamw.cu
[58/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/norm.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/norm.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/norm.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/norm.cu
[59/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/opt-step-sgd.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/opt-step-sgd.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/opt-step-sgd.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/opt-step-sgd.cu
[60/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/out-prod.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/out-prod.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/out-prod.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/out-prod.cu
[61/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/pad.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/pad.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/pad.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/pad.cu
[62/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/pad_reflect_1d.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/pad_reflect_1d.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/pad_reflect_1d.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/pad_reflect_1d.cu
[63/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/pool2d.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/pool2d.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/pool2d.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/pool2d.cu
[64/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/quantize.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/quantize.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/quantize.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:315:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  315 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:315:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  315 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:315:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  315 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:315:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  315 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:315:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  315 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:315:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  315 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:315:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  315 |             break;
      |             ^~~~~
16 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:315:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  315 |             break;
      |             ^~~~~
16 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:315:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  315 |             break;
      |             ^~~~~
16 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:315:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  315 |             break;
      |             ^~~~~
16 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:315:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  315 |             break;
      |             ^~~~~
16 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:315:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  315 |             break;
      |             ^~~~~
16 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:315:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  315 |             break;
      |             ^~~~~
16 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:315:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  315 |             break;
      |             ^~~~~
16 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:315:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  315 |             break;
      |             ^~~~~
14 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:315:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  315 |             break;
      |             ^~~~~
14 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:315:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  315 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:315:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  315 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:315:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  315 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:315:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  315 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:315:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  315 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:315:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  315 |             break;
      |             ^~~~~
17 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:315:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  315 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cu:315:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  315 |             break;
      |             ^~~~~
12 warnings generated when compiling for host.
[65/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/roll.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/roll.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/roll.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/roll.cu
[66/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/scale.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/scale.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/scale.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/scale.cu
[67/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/rope.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/rope.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/rope.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/rope.cu
[68/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/set-rows.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/set-rows.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/set-rows.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/set-rows.cu
[69/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/set.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/set.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/set.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/set.cu
[70/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/softcap.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/softcap.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/softcap.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softcap.cu
[71/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/softmax.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/softmax.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/softmax.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:184:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  184 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:221:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  221 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
2 warnings generated when compiling for gfx1010.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:184:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  184 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:221:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  221 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
2 warnings generated when compiling for gfx1012.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:184:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  184 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:221:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  221 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
2 warnings generated when compiling for gfx1030.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:184:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  184 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:221:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  221 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
2 warnings generated when compiling for gfx1031.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:184:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  184 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:221:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  221 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
2 warnings generated when compiling for gfx1035.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:184:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  184 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:221:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  221 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
2 warnings generated when compiling for gfx1036.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:184:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  184 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:221:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  221 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
2 warnings generated when compiling for gfx1100.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:184:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  184 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:221:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  221 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
2 warnings generated when compiling for gfx1101.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:184:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  184 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:221:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  221 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
2 warnings generated when compiling for gfx1102.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:184:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  184 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:221:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  221 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
2 warnings generated when compiling for gfx1103.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:184:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  184 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:221:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  221 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
2 warnings generated when compiling for gfx1150.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:184:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  184 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:221:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  221 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
2 warnings generated when compiling for gfx1151.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:184:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  184 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:221:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  221 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
2 warnings generated when compiling for gfx1152.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:184:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  184 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:221:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  221 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
2 warnings generated when compiling for gfx1153.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:184:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  184 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:221:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  221 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
2 warnings generated when compiling for gfx1200.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:184:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  184 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:221:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  221 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
2 warnings generated when compiling for gfx1201.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:184:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  184 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:221:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  221 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
2 warnings generated when compiling for gfx900.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:184:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  184 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:221:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  221 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
2 warnings generated when compiling for gfx906.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:184:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  184 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:221:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  221 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
2 warnings generated when compiling for gfx908.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:184:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  184 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:221:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  221 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
2 warnings generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:184:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  184 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:221:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  221 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
2 warnings generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:184:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  184 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:221:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  221 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
2 warnings generated when compiling for gfx942.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:184:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  184 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:221:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  221 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
2 warnings generated when compiling for gfx950.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:184:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  184 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/softmax.cu:221:13: warning: comparison of integers of different signs: 'const int' and 'unsigned int' [-Wsign-compare]
  221 |     if (tid < gridDim.x) {
      |         ~~~ ^ ~~~~~~~~~
2 warnings generated when compiling for host.
[72/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/solve_tri.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/solve_tri.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/solve_tri.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/solve_tri.cu
[73/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/ssm-conv.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/ssm-conv.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/ssm-conv.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ssm-conv.cu
[74/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/sum.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/sum.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/sum.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/sum.cu
[75/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmvf.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmvf.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmvf.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvf.cu
[76/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/sumrows.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/sumrows.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/sumrows.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/sumrows.cu
[77/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/ssm-scan.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/ssm-scan.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/ssm-scan.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/ssm-scan.cu
[78/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/top-k.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/top-k.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/top-k.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/top-k.cu
[79/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/tri.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/tri.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/tri.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/tri.cu
[80/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/tsembd.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/tsembd.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/tsembd.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/tsembd.cu
[81/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/topk-moe.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/topk-moe.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/topk-moe.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/topk-moe.cu
[82/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/upscale.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/upscale.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/upscale.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:158:89: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  158 | static __device__ float weight1(float x) { return ((a + 2) * x - (a + 3)) * x * x + 1; };
      |                                                                                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:159:95: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  159 | static __device__ float weight2(float x) { return ((a * x - 5 * a) * x + 8 * a) * x - 4 * a; };
      |                                                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:167:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  167 | };
      |  ^
3 warnings generated when compiling for gfx1010.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:158:89: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  158 | static __device__ float weight1(float x) { return ((a + 2) * x - (a + 3)) * x * x + 1; };
      |                                                                                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:159:95: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  159 | static __device__ float weight2(float x) { return ((a * x - 5 * a) * x + 8 * a) * x - 4 * a; };
      |                                                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:167:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  167 | };
      |  ^
3 warnings generated when compiling for gfx1012.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:158:89: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  158 | static __device__ float weight1(float x) { return ((a + 2) * x - (a + 3)) * x * x + 1; };
      |                                                                                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:159:95: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  159 | static __device__ float weight2(float x) { return ((a * x - 5 * a) * x + 8 * a) * x - 4 * a; };
      |                                                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:167:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  167 | };
      |  ^
3 warnings generated when compiling for gfx1030.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:158:89: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  158 | static __device__ float weight1(float x) { return ((a + 2) * x - (a + 3)) * x * x + 1; };
      |                                                                                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:159:95: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  159 | static __device__ float weight2(float x) { return ((a * x - 5 * a) * x + 8 * a) * x - 4 * a; };
      |                                                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:167:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  167 | };
      |  ^
3 warnings generated when compiling for gfx1031.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:158:89: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  158 | static __device__ float weight1(float x) { return ((a + 2) * x - (a + 3)) * x * x + 1; };
      |                                                                                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:159:95: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  159 | static __device__ float weight2(float x) { return ((a * x - 5 * a) * x + 8 * a) * x - 4 * a; };
      |                                                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:167:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  167 | };
      |  ^
3 warnings generated when compiling for gfx1035.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:158:89: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  158 | static __device__ float weight1(float x) { return ((a + 2) * x - (a + 3)) * x * x + 1; };
      |                                                                                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:159:95: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  159 | static __device__ float weight2(float x) { return ((a * x - 5 * a) * x + 8 * a) * x - 4 * a; };
      |                                                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:167:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  167 | };
      |  ^
3 warnings generated when compiling for gfx1036.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:158:89: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  158 | static __device__ float weight1(float x) { return ((a + 2) * x - (a + 3)) * x * x + 1; };
      |                                                                                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:159:95: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  159 | static __device__ float weight2(float x) { return ((a * x - 5 * a) * x + 8 * a) * x - 4 * a; };
      |                                                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:167:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  167 | };
      |  ^
3 warnings generated when compiling for gfx1100.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:158:89: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  158 | static __device__ float weight1(float x) { return ((a + 2) * x - (a + 3)) * x * x + 1; };
      |                                                                                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:159:95: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  159 | static __device__ float weight2(float x) { return ((a * x - 5 * a) * x + 8 * a) * x - 4 * a; };
      |                                                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:167:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  167 | };
      |  ^
3 warnings generated when compiling for gfx1101.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:158:89: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  158 | static __device__ float weight1(float x) { return ((a + 2) * x - (a + 3)) * x * x + 1; };
      |                                                                                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:159:95: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  159 | static __device__ float weight2(float x) { return ((a * x - 5 * a) * x + 8 * a) * x - 4 * a; };
      |                                                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:167:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  167 | };
      |  ^
3 warnings generated when compiling for gfx1102.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:158:89: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  158 | static __device__ float weight1(float x) { return ((a + 2) * x - (a + 3)) * x * x + 1; };
      |                                                                                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:159:95: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  159 | static __device__ float weight2(float x) { return ((a * x - 5 * a) * x + 8 * a) * x - 4 * a; };
      |                                                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:167:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  167 | };
      |  ^
3 warnings generated when compiling for gfx1103.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:158:89: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  158 | static __device__ float weight1(float x) { return ((a + 2) * x - (a + 3)) * x * x + 1; };
      |                                                                                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:159:95: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  159 | static __device__ float weight2(float x) { return ((a * x - 5 * a) * x + 8 * a) * x - 4 * a; };
      |                                                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:167:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  167 | };
      |  ^
3 warnings generated when compiling for gfx1150.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:158:89: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  158 | static __device__ float weight1(float x) { return ((a + 2) * x - (a + 3)) * x * x + 1; };
      |                                                                                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:159:95: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  159 | static __device__ float weight2(float x) { return ((a * x - 5 * a) * x + 8 * a) * x - 4 * a; };
      |                                                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:167:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  167 | };
      |  ^
3 warnings generated when compiling for gfx1151.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:158:89: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  158 | static __device__ float weight1(float x) { return ((a + 2) * x - (a + 3)) * x * x + 1; };
      |                                                                                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:159:95: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  159 | static __device__ float weight2(float x) { return ((a * x - 5 * a) * x + 8 * a) * x - 4 * a; };
      |                                                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:167:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  167 | };
      |  ^
3 warnings generated when compiling for gfx1152.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:158:89: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  158 | static __device__ float weight1(float x) { return ((a + 2) * x - (a + 3)) * x * x + 1; };
      |                                                                                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:159:95: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  159 | static __device__ float weight2(float x) { return ((a * x - 5 * a) * x + 8 * a) * x - 4 * a; };
      |                                                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:167:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  167 | };
      |  ^
3 warnings generated when compiling for gfx1153.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:158:89: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  158 | static __device__ float weight1(float x) { return ((a + 2) * x - (a + 3)) * x * x + 1; };
      |                                                                                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:159:95: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  159 | static __device__ float weight2(float x) { return ((a * x - 5 * a) * x + 8 * a) * x - 4 * a; };
      |                                                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:167:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  167 | };
      |  ^
3 warnings generated when compiling for gfx1200.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:158:89: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  158 | static __device__ float weight1(float x) { return ((a + 2) * x - (a + 3)) * x * x + 1; };
      |                                                                                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:159:95: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  159 | static __device__ float weight2(float x) { return ((a * x - 5 * a) * x + 8 * a) * x - 4 * a; };
      |                                                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:167:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  167 | };
      |  ^
3 warnings generated when compiling for gfx1201.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:158:89: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  158 | static __device__ float weight1(float x) { return ((a + 2) * x - (a + 3)) * x * x + 1; };
      |                                                                                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:159:95: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  159 | static __device__ float weight2(float x) { return ((a * x - 5 * a) * x + 8 * a) * x - 4 * a; };
      |                                                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:167:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  167 | };
      |  ^
3 warnings generated when compiling for gfx900.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:158:89: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  158 | static __device__ float weight1(float x) { return ((a + 2) * x - (a + 3)) * x * x + 1; };
      |                                                                                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:159:95: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  159 | static __device__ float weight2(float x) { return ((a * x - 5 * a) * x + 8 * a) * x - 4 * a; };
      |                                                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:167:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  167 | };
      |  ^
3 warnings generated when compiling for gfx906.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:158:89: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  158 | static __device__ float weight1(float x) { return ((a + 2) * x - (a + 3)) * x * x + 1; };
      |                                                                                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:159:95: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  159 | static __device__ float weight2(float x) { return ((a * x - 5 * a) * x + 8 * a) * x - 4 * a; };
      |                                                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:167:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  167 | };
      |  ^
3 warnings generated when compiling for gfx908.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:158:89: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  158 | static __device__ float weight1(float x) { return ((a + 2) * x - (a + 3)) * x * x + 1; };
      |                                                                                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:159:95: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  159 | static __device__ float weight2(float x) { return ((a * x - 5 * a) * x + 8 * a) * x - 4 * a; };
      |                                                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:167:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  167 | };
      |  ^
3 warnings generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:158:89: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  158 | static __device__ float weight1(float x) { return ((a + 2) * x - (a + 3)) * x * x + 1; };
      |                                                                                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:159:95: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  159 | static __device__ float weight2(float x) { return ((a * x - 5 * a) * x + 8 * a) * x - 4 * a; };
      |                                                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:167:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  167 | };
      |  ^
3 warnings generated when compiling for gfx90a.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:158:89: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  158 | static __device__ float weight1(float x) { return ((a + 2) * x - (a + 3)) * x * x + 1; };
      |                                                                                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:159:95: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  159 | static __device__ float weight2(float x) { return ((a * x - 5 * a) * x + 8 * a) * x - 4 * a; };
      |                                                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:167:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  167 | };
      |  ^
3 warnings generated when compiling for gfx942.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:158:89: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  158 | static __device__ float weight1(float x) { return ((a + 2) * x - (a + 3)) * x * x + 1; };
      |                                                                                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:159:95: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  159 | static __device__ float weight2(float x) { return ((a * x - 5 * a) * x + 8 * a) * x - 4 * a; };
      |                                                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:167:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  167 | };
      |  ^
3 warnings generated when compiling for gfx950.
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:158:89: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  158 | static __device__ float weight1(float x) { return ((a + 2) * x - (a + 3)) * x * x + 1; };
      |                                                                                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:159:95: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  159 | static __device__ float weight2(float x) { return ((a * x - 5 * a) * x + 8 * a) * x - 4 * a; };
      |                                                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/upscale.cu:167:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  167 | };
      |  ^
3 warnings generated when compiling for host.
[83/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/unary.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/unary.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/unary.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/unary.cu
[84/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmvq.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmvq.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmvq.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:634:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  634 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:634:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  634 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:634:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  634 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:634:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  634 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:634:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  634 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:634:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  634 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:634:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  634 |             break;
      |             ^~~~~
16 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:634:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  634 |             break;
      |             ^~~~~
16 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:634:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  634 |             break;
      |             ^~~~~
16 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:634:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  634 |             break;
      |             ^~~~~
16 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:634:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  634 |             break;
      |             ^~~~~
16 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:634:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  634 |             break;
      |             ^~~~~
16 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:634:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  634 |             break;
      |             ^~~~~
16 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:634:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  634 |             break;
      |             ^~~~~
16 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:634:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  634 |             break;
      |             ^~~~~
14 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:634:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  634 |             break;
      |             ^~~~~
14 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:634:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  634 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:634:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  634 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:634:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  634 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:634:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  634 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:634:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  634 |             break;
      |             ^~~~~
15 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:634:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  634 |             break;
      |             ^~~~~
17 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:634:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  634 |             break;
      |             ^~~~~
12 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/quantize.cuh:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/mmvq.cu:634:13: warning: 'break' will never be executed [-Wunreachable-code-break]
  634 |             break;
      |             ^~~~~
12 warnings generated when compiling for host.
[85/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/wkv.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/wkv.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/wkv.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/wkv.cu
[86/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for host.
[87/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for host.
[88/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for host.
[89/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for host.
[90/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for host.
[91/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for host.
[92/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
20 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for host.
[93/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
20 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for host.
[94/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for host.
[95/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
20 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for host.
[96/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 16, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 16, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 16, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 16, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 16, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 16, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 16, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 16, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 16, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 16, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 16, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 16, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 16, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 16, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 16, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 16, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 16, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 16, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 16, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 16, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 16, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 16, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 16, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 16, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 16, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 16, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 16, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 16, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 16, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 16, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 16, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 16, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 16, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 16, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 16, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 16, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 16, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 16, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 16, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 16, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 16, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 16, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 16, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 16, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 16, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 16, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 16, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 16, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 16, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 16, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 16, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 16, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 16, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 16, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 16, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 16, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 16, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 16, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 16, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 16, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 16, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 16, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 16, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 16, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 16, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 16, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 16, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 16, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 16, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 16, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 16, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 16, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 16, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 16, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 16, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 16, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 16, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 16, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 16, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 16, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 16, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 16, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 16, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 16, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 16, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 16, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 16, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 16, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 16, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 16, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 16, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 16, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 16, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 16, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 16, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 16, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 16, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 16, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 16, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 16, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 16, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 16, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 16, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 16, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 16, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 16, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 16, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 16, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 16, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 16, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 16, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 16, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 16, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 16, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 16, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 16, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 16, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 16, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 16, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 16, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 16, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 16, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 16, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 16, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 16, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 16, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 16, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 16, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 16, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 16, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 16, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 16, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 16, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 16, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 16, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 16, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 16, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 16, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 16, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 16, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 16, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 16, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 16, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 16, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
168 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 16, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 16, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 16, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 16, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 16, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 16, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 16, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 16, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 16, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 16, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 16, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 16, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 16, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 16, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 16, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 16, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 16, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 16, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 16, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 16, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 16, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 16, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 16, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 16, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 16, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 16, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 16, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 16, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 16, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 16, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 16, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 16, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 16, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 16, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 16, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 16, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 16, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 16, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 16, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 16, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 16, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 16, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 16, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 16, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 16, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 16, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 16, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 16, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 16, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 16, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 16, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 16, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 16, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 16, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 16, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 16, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 16, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 16, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 16, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 16, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 16, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 16, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 16, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 16, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 16, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 16, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 16, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 16, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 16, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 16, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 16, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 16, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 16, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 16, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 16, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 16, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 16, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 16, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 16, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 16, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 16, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 16, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 16, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 16, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 16, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 16, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 16, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 16, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 16, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 16, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 16, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 16, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 16, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 16, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 16, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 16, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 16, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 16, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 16, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 16, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 16, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 16, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 16, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 16, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 16, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 16, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 16, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 16, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 16, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 16, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 16, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 16, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 16, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 16, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 16, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 16, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 16, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 16, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 16, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 16, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 16, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 16, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 16, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 16, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 16, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 16, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 16, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 16, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 16, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 16, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 16, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 16, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 16, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 16, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 16, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 16, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 16, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 16, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 16, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 16, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 16, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 16, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 16, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 16, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 16, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 16, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
168 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
20 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for host.
[97/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for host.
[98/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
20 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for host.
[99/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
20 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for host.
[100/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
20 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for host.
[101/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-tile.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-wmma-f16.cuh:26:55: warning: unused parameter 'cc' [-Wunused-parameter]
   26 | static bool ggml_cuda_should_use_wmma_fattn(const int cc) {
      |                                                       ^
1 warning generated when compiling for host.
[102/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
20 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for host.
[103/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
20 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for host.
[104/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 32, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 32, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 32, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 32, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 32, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 32, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 32, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 32, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 32, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 32, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 32, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 32, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 32, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 32, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 32, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 32, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 32, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 32, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 32, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 32, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 32, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 32, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 32, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 32, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 32, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 32, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 32, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 32, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 32, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 32, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 32, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 32, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 32, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 32, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 32, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 32, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 32, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 32, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 32, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 32, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 32, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 32, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 32, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 32, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 32, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 32, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 32, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 32, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 32, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 32, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 32, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 32, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 32, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 32, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 32, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 32, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 32, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 32, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 32, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 32, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 32, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 32, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 32, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 32, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 32, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 32, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 32, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 32, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 32, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 32, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 32, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 32, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 32, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 32, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 32, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 32, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 32, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 32, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 32, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 32, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 32, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 32, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 32, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 32, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 32, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 32, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 32, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 32, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 32, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 32, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 32, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 32, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 32, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 32, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 32, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 32, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 32, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 32, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 32, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 32, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 32, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 32, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 32, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 32, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 32, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 32, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 32, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 32, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 32, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 32, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 32, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 32, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 32, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 32, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 32, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 32, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 32, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 32, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 32, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 32, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 32, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 32, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 32, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 32, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 32, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 32, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 32, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 32, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 32, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 32, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 32, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 32, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 32, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 32, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 32, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 32, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 32, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 32, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 32, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 32, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 32, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 32, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 32, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 32, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
168 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 32, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 32, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 32, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 32, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 32, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 32, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 32, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 32, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 32, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 32, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 32, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 32, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 32, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 32, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 32, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 32, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 32, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 32, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 32, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 32, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 32, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 32, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 32, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 32, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 32, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 32, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 32, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 32, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 32, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 32, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 32, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 32, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 32, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 32, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 32, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 32, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 32, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 32, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 32, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 32, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 32, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 32, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 32, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 32, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 32, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 32, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 32, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 32, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 32, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 32, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 32, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 32, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 32, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 32, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 32, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 32, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 32, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 32, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 32, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 32, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 32, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 32, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 32, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 32, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 32, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 32, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 32, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 32, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 32, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 32, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 32, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 32, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 32, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 32, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 32, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 32, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 32, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 32, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 32, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 32, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 32, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 32, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 32, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 32, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 32, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 32, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 32, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 32, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 32, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 32, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 32, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 32, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 32, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 32, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 32, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 32, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 32, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 32, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 32, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 32, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 32, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 32, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 32, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 32, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 32, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 32, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 32, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 32, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 32, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 32, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 32, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 32, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 32, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 32, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 32, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 32, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 32, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 32, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 32, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 32, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 32, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 32, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 32, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 32, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 32, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 32, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 32, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 32, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 32, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 32, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 32, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 32, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 32, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 32, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 32, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 32, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 32, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 32, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 32, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 32, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 32, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 32, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 32, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 32, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 32, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 32, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
168 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
20 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for host.
[105/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
20 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for host.
[106/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
20 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for host.
[107/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
20 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for host.
[108/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
20 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for host.
[109/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
20 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for host.
[110/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 64, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 64, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 64, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 64, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 64, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 64, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 64, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 64, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 64, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 64, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 64, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 64, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 64, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 64, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 64, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 64, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 64, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 64, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 64, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 64, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 64, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 64, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 64, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 64, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 64, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 64, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 64, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 64, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 64, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 64, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 64, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 64, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 64, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 64, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 64, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 64, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 64, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 64, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 64, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 64, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 64, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 64, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 64, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 64, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 64, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 64, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 64, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 64, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 64, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 64, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 64, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 64, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 64, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 64, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 64, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 64, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 64, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 64, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 64, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 64, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 64, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 64, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 64, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 64, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 64, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 64, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 64, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 64, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 64, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 64, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 64, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 64, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 64, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 64, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 64, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 64, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 64, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 64, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 64, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 64, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 64, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 64, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 64, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 64, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 64, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 64, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 64, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 64, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 64, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 64, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 64, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 64, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 64, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 64, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 64, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 64, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 64, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 64, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 64, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 64, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 64, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 64, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 64, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 64, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 64, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 64, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 64, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 64, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 64, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 64, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 64, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 64, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 64, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 64, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 64, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 64, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 64, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 64, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 64, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 64, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 64, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 64, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 64, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 64, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 64, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 64, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 64, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 64, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 64, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 64, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 64, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 64, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 64, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 64, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 64, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 64, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 64, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 64, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 64, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 64, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 64, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 64, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 64, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 64, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
168 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 64, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 64, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 64, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 64, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 64, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 64, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 64, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 64, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 64, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 64, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 64, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 64, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 64, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 64, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 64, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 64, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 64, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 64, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 64, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 64, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 64, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 64, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 64, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 64, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 64, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 64, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 64, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 64, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 64, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 64, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 64, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 64, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 64, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 64, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 64, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 64, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 64, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 64, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 64, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 64, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 64, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 64, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 64, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 64, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 64, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 64, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 64, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 64, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 64, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 64, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 64, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 64, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 64, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 64, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 64, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 64, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 64, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 64, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 64, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 64, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 64, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 64, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 64, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 64, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 64, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 64, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 64, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 64, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 64, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 64, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 64, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 64, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 64, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 64, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 64, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 64, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 64, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 64, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 64, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 64, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 64, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 64, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 64, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 64, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 64, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 64, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 64, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 64, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 64, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 64, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 64, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 64, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 64, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 64, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 64, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 64, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 64, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 64, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 64, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 64, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 64, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 64, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 64, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 64, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 64, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 64, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 64, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 64, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 64, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 64, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 64, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 64, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 64, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 64, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 64, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 64, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 64, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 64, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 64, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 64, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 64, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 64, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 64, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 64, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 64, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 64, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 64, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 64, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 64, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 64, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 64, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 64, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 64, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 64, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 64, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 64, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 64, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 64, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 64, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 64, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 64, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 64, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 64, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 64, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 64, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 64, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
168 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
20 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for host.
[111/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 8, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 8, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 8, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 8, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 8, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 8, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 8, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 8, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 8, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 8, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 8, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 8, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 8, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 8, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 8, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 8, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 8, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 8, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 8, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 8, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 8, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 8, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 8, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 8, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 8, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 8, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 8, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 8, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 8, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 8, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 8, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 8, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 8, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 8, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 8, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 8, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 8, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 8, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 8, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 8, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 8, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 8, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 8, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 8, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 8, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 8, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 8, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 8, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 8, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 8, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 8, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 8, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 8, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 8, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 8, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 8, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 8, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 8, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 8, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 8, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 8, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 8, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 8, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 8, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 8, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 8, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 8, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 8, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 8, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 8, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 8, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 8, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 8, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 8, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 8, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 8, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 8, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 8, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 8, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 8, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 8, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 8, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 8, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 8, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 8, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 8, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 8, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 8, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 8, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 8, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 8, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 8, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 8, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 8, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 8, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 8, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 8, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 8, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 8, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 8, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 8, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 8, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 8, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 8, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 8, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 8, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 8, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 8, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 8, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 8, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 8, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 8, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 8, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 8, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 8, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 8, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 8, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 8, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 8, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 8, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 8, 1, 2, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 8, 1, 2, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 8, 1, 2, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 8, 1, 2, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 8, 1, 2, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 8, 1, 2, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 8, 1, 2, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 8, 1, 2, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 8, 1, 2, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 8, 1, 2, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 8, 1, 2, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 8, 1, 2, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 8, 1, 2, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 8, 1, 2, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 8, 1, 2, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 8, 1, 2, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 8, 1, 2, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 8, 1, 2, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 8, 1, 2, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 8, 1, 2, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 8, 1, 2, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 8, 1, 2, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 8, 1, 2, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 8, 1, 2, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
168 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 8, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 8, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 8, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 8, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 8, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 8, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 8, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 8, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 8, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 8, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 8, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 8, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 8, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 8, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 8, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 8, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 8, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 8, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 8, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 8, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 8, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 8, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<64, 64, 8, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<64, 64, 8, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<64, 64, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 8, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 8, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 8, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 8, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 8, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 8, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 8, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 8, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 8, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 8, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 8, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 8, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 8, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 8, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 8, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 8, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 8, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 8, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 8, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 8, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 8, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 8, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<80, 80, 8, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<80, 80, 8, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<80, 80, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 8, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 8, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 8, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 8, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 8, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 8, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 8, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 8, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 8, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 8, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 8, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 8, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 8, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 8, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 8, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 8, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 8, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 8, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 8, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 8, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 8, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 8, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<96, 96, 8, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<96, 96, 8, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<96, 96, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 8, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 8, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 8, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 8, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 8, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 8, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 8, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 8, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 8, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 8, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 8, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 8, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 8, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 8, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 8, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 8, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 8, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 8, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 8, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 8, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 8, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 8, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<112, 112, 8, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<112, 112, 8, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<112, 112, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 8, 1, 4, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 8, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 8, 1, 4, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 8, 1, 4, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 8, 1, 4, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 8, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 8, 1, 4, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 8, 1, 4, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 8, 1, 4, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 8, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 8, 1, 4, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 8, 1, 4, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 8, 1, 4, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 8, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 8, 1, 4, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 8, 1, 4, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 8, 1, 4, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 8, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 8, 1, 4, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 8, 1, 4, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 8, 1, 4, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 8, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<128, 128, 8, 1, 4, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<128, 128, 8, 1, 4, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<128, 128, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 8, 1, 2, false, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 8, 1, 2, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 8, 1, 2, false, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 8, 1, 2, false, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 8, 1, 2, false, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 8, 1, 2, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 8, 1, 2, false, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 8, 1, 2, false, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 8, 1, 2, false, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 8, 1, 2, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 8, 1, 2, false, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 8, 1, 2, false, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1674:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 8, 1, false, false>' requested here
 1674 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 8, 1, 2, true, false, false, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 8, 1, 2, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 8, 1, 2, true, false, false, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1567:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 8, 1, 2, true, false, false, false>' requested here
 1567 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 8, 1, 2, true, false, true, false, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 8, 1, 2, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 8, 1, 2, true, false, true, false, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1572:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 8, 1, 2, true, false, true, false>' requested here
 1572 |             flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1081:13: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 8, 1, 2, true, false, false, true, false, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1081 |             flash_attn_ext_f16_iter
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 8, 1, 2, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:651:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  651 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1090:9: note: in instantiation of function template specialization 'flash_attn_ext_f16_iter<256, 256, 8, 1, 2, true, false, false, true, true, true, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 16, float>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>, ggml_cuda_mma::tile<16, 8, __half2>>' requested here
 1090 |         flash_attn_ext_f16_iter
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1613:5: note: in instantiation of function template specialization 'flash_attn_ext_f16_process_tile<256, 256, 8, 1, 2, true, false, false, true>' requested here
 1613 |     flash_attn_ext_f16_process_tile<DKQ, DV, ncols1, ncols2, nwarps, use_logit_softcap, V_is_K_view, needs_fixup, is_fixup>
      |     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1685:24: note: in instantiation of function template specialization 'flash_attn_ext_f16<256, 256, 8, 1, true, false>' requested here
 1685 |         fattn_kernel = flash_attn_ext_f16<DKQ, DV, ncols1, ncols2, use_logit_softcap, V_is_K_view>;
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:689:88: warning: comparison of integers of different signs: 'unsigned int' and 'const int' [-Wsign-compare]
  689 |                 if (!oob_check || k0 + (threadIdx.y % np)*T_C_KQ::J + T_C_KQ::get_j(l) < k_VKQ_sup) {
      |                                   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~
168 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
20 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for host.
[112/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
20 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for host.
[113/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
20 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for host.
[114/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
20 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for host.
[115/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
19 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:406:43: warning: unused parameter 'Q_f2' [-Wunused-parameter]
  406 |         const float2 * const __restrict__ Q_f2,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:410:43: warning: unused parameter 'dstk' [-Wunused-parameter]
  410 |         float2       * const __restrict__ dstk,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:411:43: warning: unused parameter 'dstk_fixup' [-Wunused-parameter]
  411 |         float2       * const __restrict__ dstk_fixup,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:412:21: warning: unused parameter 'scale' [-Wunused-parameter]
  412 |         const float scale,
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:416:19: warning: unused parameter 'ne02' [-Wunused-parameter]
  416 |         const int ne02,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1476:23: warning: unused parameter 'ne00' [-Wunused-parameter]
 1476 |         const int32_t ne00, const uint3   ne01, const int32_t ne02, const int32_t ne03,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:23: warning: unused parameter 'ne10' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1478:83: warning: unused parameter 'ne13' [-Wunused-parameter]
 1478 |         const int32_t ne10, const int32_t ne11, const int32_t ne12, const int32_t ne13,
      |                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:43: warning: unused parameter 'ne31' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1481:63: warning: unused parameter 'ne32' [-Wunused-parameter]
 1481 |                             const int32_t ne31, const int32_t ne32, const int32_t ne33,
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:1482:63: warning: unused parameter 'nb32' [-Wunused-parameter]
 1482 |                             const int32_t nb31, const int32_t nb32, const int64_t nb33) {
      |                                                               ^
24 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
18 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
20 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../cp-async.cuh:51:60: warning: function 'cp_async_wait_all' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
   51 | static __device__ __forceinline__ void cp_async_wait_all() {
      |                                                            ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:428:19: warning: unused parameter 'jt' [-Wunused-parameter]
  428 |         const int jt,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:430:19: warning: unused parameter 'k_VKQ_sup' [-Wunused-parameter]
  430 |         const int k_VKQ_sup) {
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:937:19: warning: unused parameter 'ne11' [-Wunused-parameter]
  937 |         const int ne11,
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../fattn-mma-f16.cuh:944:19: warning: unused parameter 'zt_gqa' [-Wunused-parameter]
  944 |         const int zt_gqa,
      |                   ^
16 warnings generated when compiling for host.
[116/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --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 ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq1_s.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq1_s.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3348:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3348 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ1_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ1_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3348:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3348 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ1_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ1_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3348:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3348 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ1_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ1_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3348:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3348 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ1_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ1_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3348:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3348 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ1_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ1_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3348:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3348 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ1_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ1_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3348:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3348 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ1_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ1_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3348:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3348 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ1_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ1_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3348:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3348 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ1_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ1_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3348:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3348 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ1_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ1_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1192:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1192 |         acc[0] = __builtin_amdgcn_mfma_i32_16x16x32_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3348:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3348 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ1_S, 8, false, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ1_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ1_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1193:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1193 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
18 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for host.
[117/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --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 ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_s.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_s.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3324:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3324 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3324:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3324 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3324:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3324 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3324:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3324 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3324:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3324 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3324:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3324 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3324:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3324 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3324:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3324 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1349:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1349 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3324:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3324 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1350:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1350 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1349:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1349 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3324:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3324 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1350:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1350 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1192:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1192 |         acc[0] = __builtin_amdgcn_mfma_i32_16x16x32_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1311:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1311 |                 mma(C, A[n], B[0]);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3324:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3324 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_S, 8, false, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1193:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1193 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
18 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for host.
[118/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --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 ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3316:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3316 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_XS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_XS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3316:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3316 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_XS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_XS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3316:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3316 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_XS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_XS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3316:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3316 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_XS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_XS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3316:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3316 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_XS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_XS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3316:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3316 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_XS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_XS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3316:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3316 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_XS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_XS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3316:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3316 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_XS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_XS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1349:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1349 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3316:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3316 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_XS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_XS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1350:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1350 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1349:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1349 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3316:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3316 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_XS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_XS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1350:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1350 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1192:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1192 |         acc[0] = __builtin_amdgcn_mfma_i32_16x16x32_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1311:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1311 |                 mma(C, A[n], B[0]);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3316:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3316 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_XS, 8, false, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_XS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1193:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1193 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
18 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for host.
[119/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --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 ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3308:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3308 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_XXS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_XXS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3308:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3308 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_XXS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_XXS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3308:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3308 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_XXS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_XXS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3308:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3308 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_XXS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_XXS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3308:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3308 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_XXS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_XXS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3308:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3308 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_XXS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_XXS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3308:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3308 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_XXS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_XXS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3308:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3308 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_XXS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_XXS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3308:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3308 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_XXS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_XXS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3308:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3308 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_XXS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_XXS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1192:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1192 |         acc[0] = __builtin_amdgcn_mfma_i32_16x16x32_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3308:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3308 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ2_XXS, 8, false, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ2_XXS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ2_XXS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1193:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1193 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
18 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for host.
[120/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --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 ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_s.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_s.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3340:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3340 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ3_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ3_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3340:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3340 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ3_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ3_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3340:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3340 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ3_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ3_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3340:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3340 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ3_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ3_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3340:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3340 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ3_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ3_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3340:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3340 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ3_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ3_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3340:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3340 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ3_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ3_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3340:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3340 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ3_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ3_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3340:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3340 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ3_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ3_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3340:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3340 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ3_S, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ3_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1192:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1192 |         acc[0] = __builtin_amdgcn_mfma_i32_16x16x32_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3340:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3340 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ3_S, 8, false, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ3_S, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_S, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1193:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1193 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
18 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_s.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for host.
[121/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --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 ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3356:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3356 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ4_NL, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ4_NL, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3356:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3356 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ4_NL, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ4_NL, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3356:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3356 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ4_NL, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ4_NL, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3356:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3356 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ4_NL, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ4_NL, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3356:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3356 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ4_NL, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ4_NL, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3356:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3356 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ4_NL, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ4_NL, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3356:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3356 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ4_NL, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ4_NL, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3356:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3356 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ4_NL, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ4_NL, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3356:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3356 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ4_NL, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ4_NL, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3356:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3356 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ4_NL, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ4_NL, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1192:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1192 |         acc[0] = __builtin_amdgcn_mfma_i32_16x16x32_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3356:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3356 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ4_NL, 8, false, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ4_NL, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_NL, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1193:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1193 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
18 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for host.
[122/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --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 ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3364:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3364 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ4_XS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ4_XS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3364:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3364 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ4_XS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ4_XS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3364:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3364 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ4_XS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ4_XS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3364:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3364 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ4_XS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ4_XS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3364:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3364 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ4_XS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ4_XS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3364:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3364 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ4_XS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ4_XS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3364:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3364 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ4_XS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ4_XS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3364:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3364 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ4_XS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ4_XS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3364:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3364 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ4_XS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ4_XS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3364:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3364 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ4_XS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ4_XS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1192:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1192 |         acc[0] = __builtin_amdgcn_mfma_i32_16x16x32_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3364:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3364 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ4_XS, 8, false, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ4_XS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ4_XS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1193:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1193 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
18 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for host.
[123/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --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 ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3332:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3332 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ3_XXS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ3_XXS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3332:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3332 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ3_XXS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ3_XXS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3332:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3332 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ3_XXS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ3_XXS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3332:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3332 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ3_XXS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ3_XXS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3332:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3332 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ3_XXS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ3_XXS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3332:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3332 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ3_XXS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ3_XXS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3332:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3332 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ3_XXS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ3_XXS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3332:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3332 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ3_XXS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ3_XXS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3332:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3332 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ3_XXS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ3_XXS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3332:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3332 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ3_XXS, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ3_XXS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1192:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1192 |         acc[0] = __builtin_amdgcn_mfma_i32_16x16x32_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3332:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3332 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_IQ3_XXS, 8, false, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_IQ3_XXS, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_IQ3_XXS, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1193:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1193 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
18 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for host.
[124/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-mxfp4.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-mxfp4.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-mxfp4.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3259:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3259 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_MXFP4, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_MXFP4, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_MXFP4, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3259:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3259 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_MXFP4, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_MXFP4, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_MXFP4, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3259:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3259 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_MXFP4, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_MXFP4, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_MXFP4, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3259:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3259 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_MXFP4, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_MXFP4, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_MXFP4, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3259:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3259 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_MXFP4, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_MXFP4, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_MXFP4, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3259:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3259 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_MXFP4, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_MXFP4, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_MXFP4, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3259:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3259 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_MXFP4, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_MXFP4, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_MXFP4, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3259:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3259 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_MXFP4, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_MXFP4, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_MXFP4, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3259:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3259 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_MXFP4, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_MXFP4, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_MXFP4, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3259:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3259 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_MXFP4, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_MXFP4, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_MXFP4, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1192:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1192 |         acc[0] = __builtin_amdgcn_mfma_i32_16x16x32_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3259:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3259 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_MXFP4, 8, false, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_MXFP4, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_MXFP4, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1193:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1193 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
18 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for host.
[125/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --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 ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_0.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_0.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3215:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_DS4>' requested here
 3215 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_DS4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3215:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_DS4>' requested here
 3215 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_DS4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3215:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_DS4>' requested here
 3215 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_DS4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3215:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_DS4>' requested here
 3215 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_DS4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3215:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_DS4>' requested here
 3215 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_DS4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3215:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_DS4>' requested here
 3215 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_DS4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3215:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_DS4>' requested here
 3215 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_DS4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3215:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_DS4>' requested here
 3215 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_DS4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3215:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_DS4>' requested here
 3215 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_DS4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3215:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_DS4>' requested here
 3215 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_DS4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1192:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1192 |         acc[0] = __builtin_amdgcn_mfma_i32_16x16x32_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3215:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_DS4>' requested here
 3215 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_DS4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_0, 8, false, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1193:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1193 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
18 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for host.
[126/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --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 ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q3_k.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q3_k.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3276:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3276 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q3_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q3_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3276:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3276 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q3_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q3_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3276:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3276 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q3_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q3_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3276:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3276 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q3_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q3_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3276:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3276 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q3_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q3_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3276:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3276 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q3_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q3_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3276:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3276 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q3_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q3_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3276:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3276 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q3_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q3_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1349:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1349 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3276:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3276 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q3_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q3_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1350:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1350 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1349:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1349 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1360:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1360 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3276:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3276 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q3_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q3_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1350:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1350 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1192:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1192 |         acc[0] = __builtin_amdgcn_mfma_i32_16x16x32_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1311:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1311 |                 mma(C, A[n], B[0]);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3276:54: note: in instantiation of function template specialization 'vec_dot_q8_0_16_q8_1_mma<8, 128>' requested here
 3276 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_16_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q3_K, 8, false, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q3_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q3_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1193:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1193 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
18 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for host.
[127/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --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 ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q2_k.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q2_k.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1691:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1691 |                 mma(Cm, A1, B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3268:54: note: in instantiation of function template specialization 'vec_dot_q2_K_q8_1_mma<8, 128>' requested here
 3268 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q2_K_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q2_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q2_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1691:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1691 |                 mma(Cm, A1, B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3268:54: note: in instantiation of function template specialization 'vec_dot_q2_K_q8_1_mma<8, 128>' requested here
 3268 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q2_K_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q2_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q2_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1691:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1691 |                 mma(Cm, A1, B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3268:54: note: in instantiation of function template specialization 'vec_dot_q2_K_q8_1_mma<8, 128>' requested here
 3268 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q2_K_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q2_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q2_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1691:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1691 |                 mma(Cm, A1, B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3268:54: note: in instantiation of function template specialization 'vec_dot_q2_K_q8_1_mma<8, 128>' requested here
 3268 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q2_K_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q2_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q2_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1691:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1691 |                 mma(Cm, A1, B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3268:54: note: in instantiation of function template specialization 'vec_dot_q2_K_q8_1_mma<8, 128>' requested here
 3268 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q2_K_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q2_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q2_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1691:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1691 |                 mma(Cm, A1, B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3268:54: note: in instantiation of function template specialization 'vec_dot_q2_K_q8_1_mma<8, 128>' requested here
 3268 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q2_K_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q2_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q2_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1691:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1691 |                 mma(Cm, A1, B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3268:54: note: in instantiation of function template specialization 'vec_dot_q2_K_q8_1_mma<8, 128>' requested here
 3268 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q2_K_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q2_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q2_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1691:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1691 |                 mma(Cm, A1, B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3268:54: note: in instantiation of function template specialization 'vec_dot_q2_K_q8_1_mma<8, 128>' requested here
 3268 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q2_K_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q2_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q2_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1349:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1349 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1691:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1691 |                 mma(Cm, A1, B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3268:54: note: in instantiation of function template specialization 'vec_dot_q2_K_q8_1_mma<8, 128>' requested here
 3268 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q2_K_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q2_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q2_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1350:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1350 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1349:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1349 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1691:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1691 |                 mma(Cm, A1, B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3268:54: note: in instantiation of function template specialization 'vec_dot_q2_K_q8_1_mma<8, 128>' requested here
 3268 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q2_K_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q2_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q2_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1350:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1350 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1192:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1192 |         acc[0] = __builtin_amdgcn_mfma_i32_16x16x32_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1623:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1623 |                 mma(Cm, A1, B[0]);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3268:54: note: in instantiation of function template specialization 'vec_dot_q2_K_q8_1_mma<8, 128>' requested here
 3268 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q2_K_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q2_K, 8, false, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q2_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q2_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1193:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1193 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
18 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for host.
[128/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --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 ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_1.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_1.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3223:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3223 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_1, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_1, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3223:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3223 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_1, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_1, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3223:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3223 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_1, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_1, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3223:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3223 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_1, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_1, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3223:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3223 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_1, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_1, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3223:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3223 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_1, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_1, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3223:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3223 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_1, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_1, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3223:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3223 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_1, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_1, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3223:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3223 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_1, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_1, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3223:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3223 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_1, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_1, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1192:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1192 |         acc[0] = __builtin_amdgcn_mfma_i32_16x16x32_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3223:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3223 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_1, 8, false, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_1, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_1, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1193:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1193 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
18 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for host.
[129/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --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 ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_k.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_k.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, false>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3284:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3284 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, true>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, true, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3922:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, true>' requested here
 3922 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
19 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, false>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3284:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3284 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, true>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, true, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3922:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, true>' requested here
 3922 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
19 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, false>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3284:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3284 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, true>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, true, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3922:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, true>' requested here
 3922 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
19 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, false>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3284:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3284 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, true>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, true, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3922:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, true>' requested here
 3922 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
19 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, false>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3284:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3284 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, true>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, true, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3922:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, true>' requested here
 3922 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
19 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, false>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3284:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3284 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, true>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, true, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3922:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, true>' requested here
 3922 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
19 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, false>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3284:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3284 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, true>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, true, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3922:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, true>' requested here
 3922 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
19 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, false>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3284:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3284 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, true>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, true, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3922:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, true>' requested here
 3922 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
19 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, false>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3284:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3284 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, true>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, true, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3922:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, true>' requested here
 3922 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
17 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, false>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3284:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3284 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, true>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, true, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3922:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, true>' requested here
 3922 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
17 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, false>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, true>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, true, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3922:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, true>' requested here
 3922 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
16 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, false>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, true>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, true, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3922:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, true>' requested here
 3922 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
16 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, false>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, true>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, true, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3922:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, true>' requested here
 3922 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
16 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, false>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1192:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1192 |         acc[0] = __builtin_amdgcn_mfma_i32_16x16x32_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3284:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3284 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, false, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1193:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1193 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3283:54: note: in instantiation of function template specialization 'load_tiles_q4_K<128, true>' requested here
 3283 |     static constexpr load_tiles_mmq_t load_tiles   = load_tiles_q4_K<mmq_y, need_check>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q4_K, 8, true, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3922:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q4_K, 8, true>' requested here
 3922 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q4_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
20 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for host.
[130/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --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 ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_1.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_1.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3239:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3239 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_1, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_1, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3239:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3239 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_1, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_1, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3239:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3239 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_1, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_1, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3239:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3239 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_1, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_1, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3239:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3239 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_1, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_1, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3239:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3239 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_1, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_1, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3239:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3239 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_1, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_1, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3239:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3239 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_1, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_1, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3239:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3239 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_1, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_1, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3239:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3239 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_1, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_1, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1192:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1192 |         acc[0] = __builtin_amdgcn_mfma_i32_16x16x32_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3239:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3239 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_1, 8, false, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_1, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_1, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1193:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1193 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
18 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for host.
[131/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --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 ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_0.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_0.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3231:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3231 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3231:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3231 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3231:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3231 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3231:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3231 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3231:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3231 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3231:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3231 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3231:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3231 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3231:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3231 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3231:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3231 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3231:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3231 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1192:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1192 |         acc[0] = __builtin_amdgcn_mfma_i32_16x16x32_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3231:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3231 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_0, 8, false, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1193:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1193 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
18 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for host.
[132/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --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 ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_k.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_k.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3292:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3292 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3292:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3292 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3292:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3292 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3292:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3292 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3292:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3292 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3292:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3292 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3292:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3292 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3292:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3292 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3292:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3292 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3292:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3292 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1192:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1192 |         acc[0] = __builtin_amdgcn_mfma_i32_16x16x32_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:1145:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 1145 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3292:54: note: in instantiation of function template specialization 'vec_dot_q8_1_q8_1_mma<8, 128>' requested here
 3292 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_1_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q5_K, 8, false, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q5_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q5_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1193:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1193 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
18 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for host.
[133/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_1.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_1.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_1.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(1);
      |                 ^
28 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(1);
      |                 ^
28 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(1);
      |                 ^
28 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(1);
      |                 ^
28 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(1);
      |                 ^
28 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(1);
      |                 ^
28 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(1);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(1);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(1);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(1);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(1);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(1);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(1);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(1);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(1);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(1);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(1);
      |                 ^
28 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(1);
      |                 ^
28 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(1);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(1);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(1);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(1);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 1, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 1, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 1, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
176 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(1);
      |                 ^
28 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(1);
      |                 ^
28 warnings generated when compiling for host.
[134/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --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 ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q6_k.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q6_k.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2501:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 2501 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3300:54: note: in instantiation of function template specialization 'vec_dot_q6_K_q8_1_mma<8, 128>' requested here
 3300 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q6_K_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q6_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q6_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2501:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 2501 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3300:54: note: in instantiation of function template specialization 'vec_dot_q6_K_q8_1_mma<8, 128>' requested here
 3300 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q6_K_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q6_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q6_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2501:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 2501 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3300:54: note: in instantiation of function template specialization 'vec_dot_q6_K_q8_1_mma<8, 128>' requested here
 3300 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q6_K_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q6_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q6_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2501:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 2501 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3300:54: note: in instantiation of function template specialization 'vec_dot_q6_K_q8_1_mma<8, 128>' requested here
 3300 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q6_K_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q6_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q6_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2501:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 2501 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3300:54: note: in instantiation of function template specialization 'vec_dot_q6_K_q8_1_mma<8, 128>' requested here
 3300 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q6_K_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q6_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q6_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2501:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 2501 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3300:54: note: in instantiation of function template specialization 'vec_dot_q6_K_q8_1_mma<8, 128>' requested here
 3300 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q6_K_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q6_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q6_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2501:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 2501 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3300:54: note: in instantiation of function template specialization 'vec_dot_q6_K_q8_1_mma<8, 128>' requested here
 3300 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q6_K_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q6_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q6_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1362:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1362 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2501:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
 2501 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3300:54: note: in instantiation of function template specialization 'vec_dot_q6_K_q8_1_mma<8, 128>' requested here
 3300 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q6_K_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q6_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q6_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1363:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1363 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1349:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1349 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2501:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 2501 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3300:54: note: in instantiation of function template specialization 'vec_dot_q6_K_q8_1_mma<8, 128>' requested here
 3300 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q6_K_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q6_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q6_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1350:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1350 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1349:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1349 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2501:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 2501 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3300:54: note: in instantiation of function template specialization 'vec_dot_q6_K_q8_1_mma<8, 128>' requested here
 3300 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q6_K_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q6_K, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q6_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1350:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1350 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1192:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1192 |         acc[0] = __builtin_amdgcn_mfma_i32_16x16x32_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2450:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
 2450 |                 mma(C, A[n], B[0]);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3300:54: note: in instantiation of function template specialization 'vec_dot_q6_K_q8_1_mma<8, 128>' requested here
 3300 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q6_K_q8_1_mma<mmq_x, mmq_y>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q6_K, 8, false, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q6_K, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q6_K, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1193:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1193 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
18 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for host.
[135/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_10.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_10.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_10.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(10);
      |                  ^
28 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(10);
      |                  ^
28 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(10);
      |                  ^
28 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(10);
      |                  ^
28 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(10);
      |                  ^
28 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(10);
      |                  ^
28 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(10);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(10);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(10);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(10);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(10);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(10);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(10);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(10);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(10);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(10);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(10);
      |                  ^
28 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(10);
      |                  ^
28 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(10);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(10);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(10);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(10);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 10, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 10, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 10, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
176 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(10);
      |                  ^
28 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(10);
      |                  ^
28 warnings generated when compiling for host.
[136/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_11.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_11.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_11.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(11);
      |                  ^
28 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(11);
      |                  ^
28 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(11);
      |                  ^
28 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(11);
      |                  ^
28 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(11);
      |                  ^
28 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(11);
      |                  ^
28 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(11);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(11);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(11);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(11);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(11);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(11);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(11);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(11);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(11);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(11);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(11);
      |                  ^
28 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(11);
      |                  ^
28 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(11);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(11);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(11);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(11);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 11, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 11, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 11, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
176 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(11);
      |                  ^
28 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(11);
      |                  ^
28 warnings generated when compiling for host.
[137/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_12.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_12.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_12.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(12);
      |                  ^
28 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(12);
      |                  ^
28 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(12);
      |                  ^
28 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(12);
      |                  ^
28 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(12);
      |                  ^
28 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(12);
      |                  ^
28 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(12);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(12);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(12);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(12);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(12);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(12);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(12);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(12);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(12);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(12);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(12);
      |                  ^
28 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(12);
      |                  ^
28 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(12);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(12);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(12);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(12);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 12, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 12, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 12, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
176 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(12);
      |                  ^
28 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(12);
      |                  ^
28 warnings generated when compiling for host.
[138/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --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 ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q8_0.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q8_0.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3247:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3247 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q8_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q8_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3247:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3247 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q8_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q8_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3247:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3247 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q8_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q8_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3247:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3247 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q8_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q8_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3247:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3247 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q8_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q8_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3247:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3247 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q8_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q8_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3247:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3247 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q8_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q8_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1237:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1237 |         int32x4_t * a_vec = (int32x4_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR_MIRRORED>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3247:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3247 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q8_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q8_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1238:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(4 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1238 |         int32x4_t * b_vec = (int32x4_t *) B.x;
      |                                           ^
17 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3247:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3247 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q8_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q8_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1214:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1214 |         int32x2_t * a_vec = (int32x2_t *) A.x;
      |                                           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3247:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3247 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3553:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q8_0, 8, false, false>' requested here
 3553 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q8_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1215:43: warning: cast from 'const int *' to '__attribute__((__vector_size__(2 * sizeof(int)))) int *' drops const qualifier [-Wcast-qual]
 1215 |         int32x2_t * b_vec = (int32x2_t *) B.x;
      |                                           ^
15 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
14 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:2054:31: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
 2054 |             for (int l = 0; l < sizeof(int); ++l) {
      |                             ~ ^ ~~~~~~~~~~~
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1192:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1192 |         acc[0] = __builtin_amdgcn_mfma_i32_16x16x32_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:909:17: note: in instantiation of function template specialization 'ggml_cuda_mma::mma<ggml_cuda_mma::DATA_LAYOUT_J_MAJOR, ggml_cuda_mma::DATA_LAYOUT_I_MAJOR>' requested here
  909 |                 mma(C, A[n], B);
      |                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3247:54: note: in instantiation of function template specialization 'vec_dot_q8_0_q8_1_mma<8, 128, MMQ_Q8_1_DS_LAYOUT_D4>' requested here
 3247 |     static constexpr vec_dot_mmq_t    vec_dot_mma  = vec_dot_q8_0_q8_1_mma<mmq_x, mmq_y, MMQ_Q8_1_DS_LAYOUT_D4>;
      |                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3633:9: note: in instantiation of function template specialization 'mul_mat_q_process_tile<GGML_TYPE_Q8_0, 8, false, false>' requested here
 3633 |         mul_mat_q_process_tile<type, mmq_x, need_check, fixup>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:3914:13: note: in instantiation of function template specialization 'mul_mat_q<GGML_TYPE_Q8_0, 8, false>' requested here
 3914 |             mul_mat_q<type, mmq_x, need_check><<<block_nums_xy_tiling, block_dims, nbytes_shared, stream>>>
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:4009:13: note: in instantiation of function template specialization 'launch_mul_mat_q<GGML_TYPE_Q8_0, 8>' requested here
 4009 |             launch_mul_mat_q<type,   8>(ctx, args, stream);
      |             ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1193:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1193 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
18 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmq.cuh:5:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
11 warnings generated when compiling for host.
[139/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_13.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_13.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_13.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(13);
      |                  ^
28 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(13);
      |                  ^
28 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(13);
      |                  ^
28 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(13);
      |                  ^
28 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(13);
      |                  ^
28 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(13);
      |                  ^
28 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(13);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(13);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(13);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(13);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(13);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(13);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(13);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(13);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(13);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(13);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(13);
      |                  ^
28 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(13);
      |                  ^
28 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(13);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(13);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(13);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(13);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 13, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 13, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 13, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
176 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(13);
      |                  ^
28 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(13);
      |                  ^
28 warnings generated when compiling for host.
[140/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_14.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_14.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_14.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(14);
      |                  ^
28 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(14);
      |                  ^
28 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(14);
      |                  ^
28 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(14);
      |                  ^
28 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(14);
      |                  ^
28 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(14);
      |                  ^
28 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(14);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(14);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(14);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(14);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(14);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(14);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(14);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(14);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(14);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(14);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(14);
      |                  ^
28 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(14);
      |                  ^
28 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(14);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(14);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(14);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(14);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 14, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 14, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 14, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
176 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(14);
      |                  ^
28 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(14);
      |                  ^
28 warnings generated when compiling for host.
[141/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_15.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_15.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_15.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(15);
      |                  ^
28 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(15);
      |                  ^
28 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(15);
      |                  ^
28 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(15);
      |                  ^
28 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(15);
      |                  ^
28 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(15);
      |                  ^
28 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(15);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(15);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(15);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(15);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(15);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(15);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(15);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(15);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(15);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(15);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(15);
      |                  ^
28 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(15);
      |                  ^
28 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(15);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(15);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(15);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(15);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 15, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 15, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 15, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
176 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(15);
      |                  ^
28 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(15);
      |                  ^
28 warnings generated when compiling for host.
[142/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_2.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_2.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_2.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(2);
      |                 ^
28 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(2);
      |                 ^
28 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(2);
      |                 ^
28 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(2);
      |                 ^
28 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(2);
      |                 ^
28 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(2);
      |                 ^
28 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(2);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(2);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(2);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(2);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(2);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(2);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(2);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(2);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(2);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(2);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(2);
      |                 ^
28 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(2);
      |                 ^
28 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(2);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(2);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(2);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(2);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 2, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 2, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 2, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
176 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(2);
      |                 ^
28 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(2);
      |                 ^
28 warnings generated when compiling for host.
[143/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_16.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_16.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_16.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(16);
      |                  ^
28 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(16);
      |                  ^
28 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(16);
      |                  ^
28 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(16);
      |                  ^
28 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(16);
      |                  ^
28 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(16);
      |                  ^
28 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(16);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(16);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(16);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(16);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(16);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(16);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(16);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(16);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(16);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(16);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(16);
      |                  ^
28 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(16);
      |                  ^
28 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(16);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(16);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(16);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(16);
      |                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 16, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 16, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 16, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
176 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(16);
      |                  ^
28 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu:5:18: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(16);
      |                  ^
28 warnings generated when compiling for host.
[144/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_3.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_3.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_3.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(3);
      |                 ^
28 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(3);
      |                 ^
28 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(3);
      |                 ^
28 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(3);
      |                 ^
28 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(3);
      |                 ^
28 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(3);
      |                 ^
28 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(3);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(3);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(3);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(3);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(3);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(3);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(3);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(3);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(3);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(3);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(3);
      |                 ^
28 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(3);
      |                 ^
28 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(3);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(3);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(3);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(3);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 3, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 3, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 3, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
176 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(3);
      |                 ^
28 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(3);
      |                 ^
28 warnings generated when compiling for host.
[145/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_4.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_4.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_4.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(4);
      |                 ^
28 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(4);
      |                 ^
28 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(4);
      |                 ^
28 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(4);
      |                 ^
28 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(4);
      |                 ^
28 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(4);
      |                 ^
28 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(4);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(4);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(4);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(4);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(4);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(4);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(4);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(4);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(4);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(4);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(4);
      |                 ^
28 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(4);
      |                 ^
28 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(4);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(4);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(4);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(4);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 4, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 4, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 4, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
176 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(4);
      |                 ^
28 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(4);
      |                 ^
28 warnings generated when compiling for host.
[146/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_5.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_5.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_5.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(5);
      |                 ^
28 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(5);
      |                 ^
28 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(5);
      |                 ^
28 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(5);
      |                 ^
28 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(5);
      |                 ^
28 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(5);
      |                 ^
28 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(5);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(5);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(5);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(5);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(5);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(5);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(5);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(5);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(5);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(5);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(5);
      |                 ^
28 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(5);
      |                 ^
28 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(5);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(5);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(5);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(5);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 5, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 5, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 5, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
176 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(5);
      |                 ^
28 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(5);
      |                 ^
28 warnings generated when compiling for host.
[147/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_6.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_6.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_6.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(6);
      |                 ^
28 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(6);
      |                 ^
28 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(6);
      |                 ^
28 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(6);
      |                 ^
28 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(6);
      |                 ^
28 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(6);
      |                 ^
28 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(6);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(6);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(6);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(6);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(6);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(6);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(6);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(6);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(6);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(6);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(6);
      |                 ^
28 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(6);
      |                 ^
28 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(6);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(6);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(6);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(6);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 6, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 6, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 6, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
176 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(6);
      |                 ^
28 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(6);
      |                 ^
28 warnings generated when compiling for host.
[148/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_7.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_7.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_7.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(7);
      |                 ^
28 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(7);
      |                 ^
28 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(7);
      |                 ^
28 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(7);
      |                 ^
28 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(7);
      |                 ^
28 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(7);
      |                 ^
28 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(7);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(7);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(7);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(7);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(7);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(7);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(7);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(7);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(7);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(7);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(7);
      |                 ^
28 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(7);
      |                 ^
28 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(7);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(7);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(7);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(7);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 7, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 7, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 7, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
176 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(7);
      |                 ^
28 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(7);
      |                 ^
28 warnings generated when compiling for host.
[149/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_8.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_8.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_8.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(8);
      |                 ^
28 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(8);
      |                 ^
28 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(8);
      |                 ^
28 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(8);
      |                 ^
28 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(8);
      |                 ^
28 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(8);
      |                 ^
28 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(8);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(8);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(8);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(8);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(8);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(8);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(8);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(8);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(8);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(8);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(8);
      |                 ^
28 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(8);
      |                 ^
28 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(8);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(8);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(8);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(8);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 8, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 8, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 8, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
176 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(8);
      |                 ^
28 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(8);
      |                 ^
28 warnings generated when compiling for host.
[150/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_9.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_9.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_9.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(9);
      |                 ^
28 warnings generated when compiling for gfx1010.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(9);
      |                 ^
28 warnings generated when compiling for gfx1012.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(9);
      |                 ^
28 warnings generated when compiling for gfx1030.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(9);
      |                 ^
28 warnings generated when compiling for gfx1031.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(9);
      |                 ^
28 warnings generated when compiling for gfx1035.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(9);
      |                 ^
28 warnings generated when compiling for gfx1036.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(9);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1100.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(9);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1101.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(9);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1102.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(9);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1103.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(9);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1150.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(9);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1151.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(9);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1152.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(9);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
127 warnings generated when compiling for gfx1153.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(9);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1200.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:198:63: warning: unused parameter 'l' [-Wunused-parameter]
  198 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:325:63: warning: unused parameter 'l' [-Wunused-parameter]
  325 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:682:99: warning: unused parameter 't' [-Wunused-parameter]
  682 |     static __device__ __forceinline__ tile<8, 8, half2> get_transposed(const tile<16, 4, half2> & t) {
      |                                                                                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(9);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 32, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 32, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 32, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 32, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 32, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
125 warnings generated when compiling for gfx1201.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(9);
      |                 ^
28 warnings generated when compiling for gfx900.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(9);
      |                 ^
28 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(9);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx908.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(9);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(9);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
174 warnings generated when compiling for gfx90a.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:124:63: warning: unused parameter 'l' [-Wunused-parameter]
  124 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:351:63: warning: unused parameter 'l' [-Wunused-parameter]
  351 |         static __device__ __forceinline__ int get_i(const int l) {
      |                                                               ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:113:13: warning: zero size arrays are an extension [-Wzero-length-array]
  113 |         T x[ne] = {0};
      |             ^~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:870:30: note: in instantiation of template class 'ggml_cuda_mma::tile<8, 4, int>' requested here
  870 |         GGML_UNUSED_VARS(D, A, B);
      |                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include/ggml.h:264:50: note: expanded from macro 'GGML_UNUSED_VARS'
  264 | #define GGML_UNUSED_VARS(...) do { (void)sizeof((__VA_ARGS__, 0)); } while(0)
      |                                                  ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1271:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1271 |         acc[0] = __builtin_amdgcn_mfma_i32_32x32x16_i8(((int64_t *) A.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1272:69: warning: cast from 'const int *' to 'long *' drops const qualifier [-Wcast-qual]
 1272 |                                                        ((int64_t *) B.x)[0],
      |                                                                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(9);
      |                 ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<float, 64, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<float, 64, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<float, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__half2, 64, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__half2, 64, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__half2, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 1>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 1, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 1, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:670:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 1>' requested here
  670 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 1>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 2>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 2, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 2, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:677:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 2>' requested here
  677 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 2>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 3>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 3, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 3, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:684:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 3>' requested here
  684 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 3>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 4>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 4, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 4, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:691:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 4>' requested here
  691 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 4>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 5>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 5, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 5, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:698:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 5>' requested here
  698 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 5>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 6>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 6, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 6, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:705:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 6>' requested here
  705 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 6>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 7>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 7, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 7, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:712:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 7>' requested here
  712 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 7>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:541:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  541 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:596:9: note: in instantiation of function template specialization 'mul_mat_f_ids<__hip_bfloat162, 64, 9, 8>' requested here
  596 |         mul_mat_f_ids<T, rows_per_block, cols_per_block, nwarps><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:556:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  556 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:607:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 8, true>' requested here
  607 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, true><<<block_nums_ids, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:279:37: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  279 |                 for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                                  ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:263:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  263 |             for (int i1 = 0; i1 < sizeof(sum)/sizeof(sum[0]); ++i1) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:612:9: note: in instantiation of function template specialization 'mul_mat_f<__hip_bfloat162, 64, 9, 8, false>' requested here
  612 |         mul_mat_f<T, rows_per_block, cols_per_block, nwarps, false><<<block_nums, block_dims, nbytes_shared_total, stream>>>
      |         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:719:13: note: in instantiation of function template specialization 'mul_mat_f_switch_ids<__hip_bfloat162, 64, 9, 8>' requested here
  719 |             mul_mat_f_switch_ids<T, rows_per_block, cols_per_block, 8>(
      |             ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:272:33: warning: comparison of integers of different signs: 'int' and '__size_t' (aka 'unsigned long') [-Wsign-compare]
  272 |             for (int i0 = 0; i0 < sizeof(sum)/sizeof(sum[0]); ++i0) {
      |                              ~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~
176 warnings generated when compiling for gfx942.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(9);
      |                 ^
28 warnings generated when compiling for gfx950.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:704:86: warning: function 'make_identity_mat' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  704 |     static __device__ __forceinline__ void make_identity_mat(tile<16, 8, half2> & t) {
      |                                                                                      ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:829:87: warning: function 'load_ldmatrix' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  829 |             tile<32, 4, half2> & t, const half2 * __restrict__ xs0, const int stride) {
      |                                                                                       ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:854:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  854 |             tile<16, 8, int> & D, const tile<16, 4, int> & A, const tile<8, 4, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:876:90: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  876 |             tile<16, 8, int> & D, const tile<16, 8, int> & A, const tile<8, 8, int> & B) {
      |                                                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:904:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  904 |             tile<16, 4, half2> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:929:97: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  929 |             tile<16, 8, half2> & D, const tile<16, 8, half2> & A, const tile<16, 8, half2> & B) {
      |                                                                                                 ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1037:96: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1037 |             tile<16, 8, float> & D, const tile<16, 8, half2> & A, const tile<8, 8, half2> & B) {
      |                                                                                                ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1062:110: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1062 |             tile<16, 8, float> & D, const tile<16, 8, nv_bfloat162> & A, const tile<8, 8, nv_bfloat162> & B) {
      |                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1266:92: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1266 |             tile<32, 32, int> & D, const tile<32, 4, int> & A, const tile<32, 4, int> & B) {
      |                                                                                            ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1302:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1302 |             tile<32, 8, float> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_I_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mma.cuh:1322:126: warning: function 'mma' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
 1322 |             tile<32, 4, half2> & D, const tile<32, 4, half2> & A, const tile<8, 4, half2, DATA_LAYOUT_J_MAJOR_MIRRORED> & B) {
      |                                                                                                                              ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:3:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:890:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  890 | DECL_MMF_CASE_EXTERN(1);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:891:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  891 | DECL_MMF_CASE_EXTERN(2);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:892:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  892 | DECL_MMF_CASE_EXTERN(3);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:893:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  893 | DECL_MMF_CASE_EXTERN(4);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:894:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  894 | DECL_MMF_CASE_EXTERN(5);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:895:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  895 | DECL_MMF_CASE_EXTERN(6);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:896:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  896 | DECL_MMF_CASE_EXTERN(7);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:897:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  897 | DECL_MMF_CASE_EXTERN(8);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:898:24: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  898 | DECL_MMF_CASE_EXTERN(9);
      |                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:899:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  899 | DECL_MMF_CASE_EXTERN(10);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:900:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  900 | DECL_MMF_CASE_EXTERN(11);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:901:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  901 | DECL_MMF_CASE_EXTERN(12);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:902:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  902 | DECL_MMF_CASE_EXTERN(13);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:903:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  903 | DECL_MMF_CASE_EXTERN(14);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:904:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  904 | DECL_MMF_CASE_EXTERN(15);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/../mmf.cuh:905:25: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
  905 | DECL_MMF_CASE_EXTERN(16);
      |                         ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu:5:17: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
    5 | DECL_MMF_CASE(9);
      |                 ^
28 warnings generated when compiling for host.
[151/401] : && /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  -shared -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 -Xlinker --dependency-file=ggml/src/CMakeFiles/ggml-cpu.dir/link.d -Wl,-soname,libggml-cpu.so.0 -o bin/libggml-cpu.so.0.9.5 ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu.c.o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ggml-cpu.cpp.o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/repack.cpp.o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/hbm.cpp.o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/quants.c.o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/traits.cpp.o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/amx/amx.cpp.o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/amx/mmq.cpp.o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/binary-ops.cpp.o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/unary-ops.cpp.o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/vec.cpp.o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/ops.cpp.o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/llamafile/sgemm.cpp.o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/arch/x86/quants.c.o ggml/src/CMakeFiles/ggml-cpu.dir/ggml-cpu/arch/x86/repack.cpp.o  bin/libggml-base.so.0.9.5 && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
[152/401] /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-b8064-build/llama.cpp-b8064/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-dl.cpp.o -MF ggml/src/CMakeFiles/ggml.dir/ggml-backend-dl.cpp.o.d -o ggml/src/CMakeFiles/ggml.dir/ggml-backend-dl.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-backend-dl.cpp
[153/401] /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-b8064-build/llama.cpp-b8064/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 ggml/src/CMakeFiles/ggml.dir/ggml-backend-reg.cpp.o.d -o ggml/src/CMakeFiles/ggml.dir/ggml-backend-reg.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-backend-reg.cpp
[154/401] /usr/bin/cmake -E cmake_symlink_library bin/libggml-cpu.so.0.9.5  bin/libggml-cpu.so.0 bin/libggml-cpu.so && :
[155/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 src/CMakeFiles/llama.dir/llama.cpp.o.d -o src/CMakeFiles/llama.dir/llama.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama.cpp
[156/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 src/CMakeFiles/llama.dir/llama-adapter.cpp.o.d -o src/CMakeFiles/llama.dir/llama-adapter.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama-adapter.cpp
[157/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 src/CMakeFiles/llama.dir/llama-arch.cpp.o.d -o src/CMakeFiles/llama.dir/llama-arch.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama-arch.cpp
[158/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 src/CMakeFiles/llama.dir/llama-batch.cpp.o.d -o src/CMakeFiles/llama.dir/llama-batch.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama-batch.cpp
[159/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 src/CMakeFiles/llama.dir/llama-chat.cpp.o.d -o src/CMakeFiles/llama.dir/llama-chat.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama-chat.cpp
[160/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 src/CMakeFiles/llama.dir/llama-context.cpp.o.d -o src/CMakeFiles/llama.dir/llama-context.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama-context.cpp
[161/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-cparams.cpp.o -MF src/CMakeFiles/llama.dir/llama-cparams.cpp.o.d -o src/CMakeFiles/llama.dir/llama-cparams.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama-cparams.cpp
[162/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 src/CMakeFiles/llama.dir/llama-grammar.cpp.o.d -o src/CMakeFiles/llama.dir/llama-grammar.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama-grammar.cpp
[163/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-graph.cpp.o -MF src/CMakeFiles/llama.dir/llama-graph.cpp.o.d -o src/CMakeFiles/llama.dir/llama-graph.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama-graph.cpp
[164/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 src/CMakeFiles/llama.dir/llama-hparams.cpp.o.d -o src/CMakeFiles/llama.dir/llama-hparams.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama-hparams.cpp
[165/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 src/CMakeFiles/llama.dir/llama-impl.cpp.o.d -o src/CMakeFiles/llama.dir/llama-impl.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama-impl.cpp
[166/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-io.cpp.o -MF src/CMakeFiles/llama.dir/llama-io.cpp.o.d -o src/CMakeFiles/llama.dir/llama-io.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama-io.cpp
[167/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 src/CMakeFiles/llama.dir/llama-kv-cache.cpp.o.d -o src/CMakeFiles/llama.dir/llama-kv-cache.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama-kv-cache.cpp
[168/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-iswa.cpp.o -MF src/CMakeFiles/llama.dir/llama-kv-cache-iswa.cpp.o.d -o src/CMakeFiles/llama.dir/llama-kv-cache-iswa.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama-kv-cache-iswa.cpp
[169/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-memory.cpp.o -MF src/CMakeFiles/llama.dir/llama-memory.cpp.o.d -o src/CMakeFiles/llama.dir/llama-memory.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama-memory.cpp
[170/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-memory-hybrid.cpp.o -MF src/CMakeFiles/llama.dir/llama-memory-hybrid.cpp.o.d -o src/CMakeFiles/llama.dir/llama-memory-hybrid.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama-memory-hybrid.cpp
[171/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-memory-hybrid-iswa.cpp.o -MF src/CMakeFiles/llama.dir/llama-memory-hybrid-iswa.cpp.o.d -o src/CMakeFiles/llama.dir/llama-memory-hybrid-iswa.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama-memory-hybrid-iswa.cpp
[172/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-memory-recurrent.cpp.o -MF src/CMakeFiles/llama.dir/llama-memory-recurrent.cpp.o.d -o src/CMakeFiles/llama.dir/llama-memory-recurrent.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama-memory-recurrent.cpp
[173/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 src/CMakeFiles/llama.dir/llama-mmap.cpp.o.d -o src/CMakeFiles/llama.dir/llama-mmap.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama-mmap.cpp
[174/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 src/CMakeFiles/llama.dir/llama-model-loader.cpp.o.d -o src/CMakeFiles/llama.dir/llama-model-loader.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama-model-loader.cpp
[175/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-saver.cpp.o -MF src/CMakeFiles/llama.dir/llama-model-saver.cpp.o.d -o src/CMakeFiles/llama.dir/llama-model-saver.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama-model-saver.cpp
[176/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 src/CMakeFiles/llama.dir/llama-model.cpp.o.d -o src/CMakeFiles/llama.dir/llama-model.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama-model.cpp
[177/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 src/CMakeFiles/llama.dir/llama-quant.cpp.o.d -o src/CMakeFiles/llama.dir/llama-quant.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama-quant.cpp
[178/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-sampler.cpp.o -MF src/CMakeFiles/llama.dir/llama-sampler.cpp.o.d -o src/CMakeFiles/llama.dir/llama-sampler.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama-sampler.cpp
[179/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 src/CMakeFiles/llama.dir/llama-vocab.cpp.o.d -o src/CMakeFiles/llama.dir/llama-vocab.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/llama-vocab.cpp
[180/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-instance-q8_0-q8_0.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-instance-q8_0-q8_0.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-instance-q8_0-q8_0.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q8_0-q8_0.cu
[181/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-instance-f16-f16.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-instance-f16-f16.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-instance-f16-f16.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-f16-f16.cu
[182/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 src/CMakeFiles/llama.dir/unicode-data.cpp.o.d -o src/CMakeFiles/llama.dir/unicode-data.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/unicode-data.cpp
[183/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/apertus.cpp.o -MF src/CMakeFiles/llama.dir/models/apertus.cpp.o.d -o src/CMakeFiles/llama.dir/models/apertus.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/apertus.cpp
[184/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/afmoe.cpp.o -MF src/CMakeFiles/llama.dir/models/afmoe.cpp.o.d -o src/CMakeFiles/llama.dir/models/afmoe.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/afmoe.cpp
[185/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 src/CMakeFiles/llama.dir/unicode.cpp.o.d -o src/CMakeFiles/llama.dir/unicode.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/unicode.cpp
[186/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/arcee.cpp.o -MF src/CMakeFiles/llama.dir/models/arcee.cpp.o.d -o src/CMakeFiles/llama.dir/models/arcee.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/arcee.cpp
[187/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/arctic.cpp.o -MF src/CMakeFiles/llama.dir/models/arctic.cpp.o.d -o src/CMakeFiles/llama.dir/models/arctic.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/arctic.cpp
[188/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/arwkv7.cpp.o -MF src/CMakeFiles/llama.dir/models/arwkv7.cpp.o.d -o src/CMakeFiles/llama.dir/models/arwkv7.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/arwkv7.cpp
[189/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/baichuan.cpp.o -MF src/CMakeFiles/llama.dir/models/baichuan.cpp.o.d -o src/CMakeFiles/llama.dir/models/baichuan.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/baichuan.cpp
[190/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/bailingmoe.cpp.o -MF src/CMakeFiles/llama.dir/models/bailingmoe.cpp.o.d -o src/CMakeFiles/llama.dir/models/bailingmoe.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/bailingmoe.cpp
[191/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/bailingmoe2.cpp.o -MF src/CMakeFiles/llama.dir/models/bailingmoe2.cpp.o.d -o src/CMakeFiles/llama.dir/models/bailingmoe2.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/bailingmoe2.cpp
[192/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/bert.cpp.o -MF src/CMakeFiles/llama.dir/models/bert.cpp.o.d -o src/CMakeFiles/llama.dir/models/bert.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/bert.cpp
[193/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/bitnet.cpp.o -MF src/CMakeFiles/llama.dir/models/bitnet.cpp.o.d -o src/CMakeFiles/llama.dir/models/bitnet.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/bitnet.cpp
[194/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/bloom.cpp.o -MF src/CMakeFiles/llama.dir/models/bloom.cpp.o.d -o src/CMakeFiles/llama.dir/models/bloom.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/bloom.cpp
[195/401] /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-b8064-build/llama.cpp-b8064/ggml/src/ggml-hip/.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -x c++ -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=gfx950 --offload-arch=gfx1010 --offload-arch=gfx1012 --offload-arch=gfx1030 --offload-arch=gfx1031 --offload-arch=gfx1035 --offload-arch=gfx1036 --offload-arch=gfx1100 --offload-arch=gfx1101 --offload-arch=gfx1102 --offload-arch=gfx1103 --offload-arch=gfx1150 --offload-arch=gfx1151 --offload-arch=gfx1152 --offload-arch=gfx1153 --offload-arch=gfx1200 --offload-arch=gfx1201 -MD -MT ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-instance-q4_0-q4_0.cu.o -MF ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-instance-q4_0-q4_0.cu.o.d -o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-instance-q4_0-q4_0.cu.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_0-q4_0.cu
[196/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/chameleon.cpp.o -MF src/CMakeFiles/llama.dir/models/chameleon.cpp.o.d -o src/CMakeFiles/llama.dir/models/chameleon.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/chameleon.cpp
[197/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/chatglm.cpp.o -MF src/CMakeFiles/llama.dir/models/chatglm.cpp.o.d -o src/CMakeFiles/llama.dir/models/chatglm.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/chatglm.cpp
[198/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/codeshell.cpp.o -MF src/CMakeFiles/llama.dir/models/codeshell.cpp.o.d -o src/CMakeFiles/llama.dir/models/codeshell.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/codeshell.cpp
[199/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/cogvlm.cpp.o -MF src/CMakeFiles/llama.dir/models/cogvlm.cpp.o.d -o src/CMakeFiles/llama.dir/models/cogvlm.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/cogvlm.cpp
[200/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/cohere2-iswa.cpp.o -MF src/CMakeFiles/llama.dir/models/cohere2-iswa.cpp.o.d -o src/CMakeFiles/llama.dir/models/cohere2-iswa.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/cohere2-iswa.cpp
[201/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/command-r.cpp.o -MF src/CMakeFiles/llama.dir/models/command-r.cpp.o.d -o src/CMakeFiles/llama.dir/models/command-r.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/command-r.cpp
[202/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/dbrx.cpp.o -MF src/CMakeFiles/llama.dir/models/dbrx.cpp.o.d -o src/CMakeFiles/llama.dir/models/dbrx.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/dbrx.cpp
[203/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/deci.cpp.o -MF src/CMakeFiles/llama.dir/models/deci.cpp.o.d -o src/CMakeFiles/llama.dir/models/deci.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/deci.cpp
[204/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/deepseek.cpp.o -MF src/CMakeFiles/llama.dir/models/deepseek.cpp.o.d -o src/CMakeFiles/llama.dir/models/deepseek.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/deepseek.cpp
[205/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/deepseek2.cpp.o -MF src/CMakeFiles/llama.dir/models/deepseek2.cpp.o.d -o src/CMakeFiles/llama.dir/models/deepseek2.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/deepseek2.cpp
[206/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/dots1.cpp.o -MF src/CMakeFiles/llama.dir/models/dots1.cpp.o.d -o src/CMakeFiles/llama.dir/models/dots1.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/dots1.cpp
[207/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/dream.cpp.o -MF src/CMakeFiles/llama.dir/models/dream.cpp.o.d -o src/CMakeFiles/llama.dir/models/dream.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/dream.cpp
[208/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/ernie4-5-moe.cpp.o -MF src/CMakeFiles/llama.dir/models/ernie4-5-moe.cpp.o.d -o src/CMakeFiles/llama.dir/models/ernie4-5-moe.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/ernie4-5-moe.cpp
[209/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/ernie4-5.cpp.o -MF src/CMakeFiles/llama.dir/models/ernie4-5.cpp.o.d -o src/CMakeFiles/llama.dir/models/ernie4-5.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/ernie4-5.cpp
[210/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/exaone.cpp.o -MF src/CMakeFiles/llama.dir/models/exaone.cpp.o.d -o src/CMakeFiles/llama.dir/models/exaone.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/exaone.cpp
[211/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/exaone4.cpp.o -MF src/CMakeFiles/llama.dir/models/exaone4.cpp.o.d -o src/CMakeFiles/llama.dir/models/exaone4.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/exaone4.cpp
[212/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/exaone-moe.cpp.o -MF src/CMakeFiles/llama.dir/models/exaone-moe.cpp.o.d -o src/CMakeFiles/llama.dir/models/exaone-moe.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/exaone-moe.cpp
[213/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/falcon-h1.cpp.o -MF src/CMakeFiles/llama.dir/models/falcon-h1.cpp.o.d -o src/CMakeFiles/llama.dir/models/falcon-h1.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/falcon-h1.cpp
[214/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/falcon.cpp.o -MF src/CMakeFiles/llama.dir/models/falcon.cpp.o.d -o src/CMakeFiles/llama.dir/models/falcon.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/falcon.cpp
[215/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/gemma-embedding.cpp.o -MF src/CMakeFiles/llama.dir/models/gemma-embedding.cpp.o.d -o src/CMakeFiles/llama.dir/models/gemma-embedding.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/gemma-embedding.cpp
[216/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/gemma.cpp.o -MF src/CMakeFiles/llama.dir/models/gemma.cpp.o.d -o src/CMakeFiles/llama.dir/models/gemma.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/gemma.cpp
[217/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/gemma3.cpp.o -MF src/CMakeFiles/llama.dir/models/gemma3.cpp.o.d -o src/CMakeFiles/llama.dir/models/gemma3.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/gemma3.cpp
[218/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/gemma2-iswa.cpp.o -MF src/CMakeFiles/llama.dir/models/gemma2-iswa.cpp.o.d -o src/CMakeFiles/llama.dir/models/gemma2-iswa.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/gemma2-iswa.cpp
[219/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/gemma3n-iswa.cpp.o -MF src/CMakeFiles/llama.dir/models/gemma3n-iswa.cpp.o.d -o src/CMakeFiles/llama.dir/models/gemma3n-iswa.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/gemma3n-iswa.cpp
[220/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/glm4-moe.cpp.o -MF src/CMakeFiles/llama.dir/models/glm4-moe.cpp.o.d -o src/CMakeFiles/llama.dir/models/glm4-moe.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/glm4-moe.cpp
[221/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/glm4.cpp.o -MF src/CMakeFiles/llama.dir/models/glm4.cpp.o.d -o src/CMakeFiles/llama.dir/models/glm4.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/glm4.cpp
[222/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/gpt2.cpp.o -MF src/CMakeFiles/llama.dir/models/gpt2.cpp.o.d -o src/CMakeFiles/llama.dir/models/gpt2.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/gpt2.cpp
[223/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/gptneox.cpp.o -MF src/CMakeFiles/llama.dir/models/gptneox.cpp.o.d -o src/CMakeFiles/llama.dir/models/gptneox.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/gptneox.cpp
[224/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/granite-hybrid.cpp.o -MF src/CMakeFiles/llama.dir/models/granite-hybrid.cpp.o.d -o src/CMakeFiles/llama.dir/models/granite-hybrid.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/granite-hybrid.cpp
[225/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/granite.cpp.o -MF src/CMakeFiles/llama.dir/models/granite.cpp.o.d -o src/CMakeFiles/llama.dir/models/granite.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/granite.cpp
[226/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/grok.cpp.o -MF src/CMakeFiles/llama.dir/models/grok.cpp.o.d -o src/CMakeFiles/llama.dir/models/grok.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/grok.cpp
[227/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/grovemoe.cpp.o -MF src/CMakeFiles/llama.dir/models/grovemoe.cpp.o.d -o src/CMakeFiles/llama.dir/models/grovemoe.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/grovemoe.cpp
[228/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/hunyuan-dense.cpp.o -MF src/CMakeFiles/llama.dir/models/hunyuan-dense.cpp.o.d -o src/CMakeFiles/llama.dir/models/hunyuan-dense.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/hunyuan-dense.cpp
[229/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/hunyuan-moe.cpp.o -MF src/CMakeFiles/llama.dir/models/hunyuan-moe.cpp.o.d -o src/CMakeFiles/llama.dir/models/hunyuan-moe.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/hunyuan-moe.cpp
[230/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/internlm2.cpp.o -MF src/CMakeFiles/llama.dir/models/internlm2.cpp.o.d -o src/CMakeFiles/llama.dir/models/internlm2.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/internlm2.cpp
[231/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/jais.cpp.o -MF src/CMakeFiles/llama.dir/models/jais.cpp.o.d -o src/CMakeFiles/llama.dir/models/jais.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/jais.cpp
[232/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/jamba.cpp.o -MF src/CMakeFiles/llama.dir/models/jamba.cpp.o.d -o src/CMakeFiles/llama.dir/models/jamba.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/jamba.cpp
[233/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/kimi-linear.cpp.o -MF src/CMakeFiles/llama.dir/models/kimi-linear.cpp.o.d -o src/CMakeFiles/llama.dir/models/kimi-linear.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/kimi-linear.cpp
[234/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/lfm2.cpp.o -MF src/CMakeFiles/llama.dir/models/lfm2.cpp.o.d -o src/CMakeFiles/llama.dir/models/lfm2.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/lfm2.cpp
[235/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/llada-moe.cpp.o -MF src/CMakeFiles/llama.dir/models/llada-moe.cpp.o.d -o src/CMakeFiles/llama.dir/models/llada-moe.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/llada-moe.cpp
[236/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/llama-iswa.cpp.o -MF src/CMakeFiles/llama.dir/models/llama-iswa.cpp.o.d -o src/CMakeFiles/llama.dir/models/llama-iswa.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/llama-iswa.cpp
[237/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/llada.cpp.o -MF src/CMakeFiles/llama.dir/models/llada.cpp.o.d -o src/CMakeFiles/llama.dir/models/llada.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/llada.cpp
[238/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/llama.cpp.o -MF src/CMakeFiles/llama.dir/models/llama.cpp.o.d -o src/CMakeFiles/llama.dir/models/llama.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/llama.cpp
[239/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/mamba.cpp.o -MF src/CMakeFiles/llama.dir/models/mamba.cpp.o.d -o src/CMakeFiles/llama.dir/models/mamba.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/mamba.cpp
[240/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/maincoder.cpp.o -MF src/CMakeFiles/llama.dir/models/maincoder.cpp.o.d -o src/CMakeFiles/llama.dir/models/maincoder.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/maincoder.cpp
[241/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/mimo2-iswa.cpp.o -MF src/CMakeFiles/llama.dir/models/mimo2-iswa.cpp.o.d -o src/CMakeFiles/llama.dir/models/mimo2-iswa.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/mimo2-iswa.cpp
[242/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/minicpm3.cpp.o -MF src/CMakeFiles/llama.dir/models/minicpm3.cpp.o.d -o src/CMakeFiles/llama.dir/models/minicpm3.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/minicpm3.cpp
[243/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/minimax-m2.cpp.o -MF src/CMakeFiles/llama.dir/models/minimax-m2.cpp.o.d -o src/CMakeFiles/llama.dir/models/minimax-m2.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/minimax-m2.cpp
[244/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/modern-bert.cpp.o -MF src/CMakeFiles/llama.dir/models/modern-bert.cpp.o.d -o src/CMakeFiles/llama.dir/models/modern-bert.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/modern-bert.cpp
[245/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/mpt.cpp.o -MF src/CMakeFiles/llama.dir/models/mpt.cpp.o.d -o src/CMakeFiles/llama.dir/models/mpt.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/mpt.cpp
[246/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/nemotron-h.cpp.o -MF src/CMakeFiles/llama.dir/models/nemotron-h.cpp.o.d -o src/CMakeFiles/llama.dir/models/nemotron-h.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/nemotron-h.cpp
[247/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/nemotron.cpp.o -MF src/CMakeFiles/llama.dir/models/nemotron.cpp.o.d -o src/CMakeFiles/llama.dir/models/nemotron.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/nemotron.cpp
[248/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/neo-bert.cpp.o -MF src/CMakeFiles/llama.dir/models/neo-bert.cpp.o.d -o src/CMakeFiles/llama.dir/models/neo-bert.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/neo-bert.cpp
[249/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/olmo.cpp.o -MF src/CMakeFiles/llama.dir/models/olmo.cpp.o.d -o src/CMakeFiles/llama.dir/models/olmo.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/olmo.cpp
[250/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/olmo2.cpp.o -MF src/CMakeFiles/llama.dir/models/olmo2.cpp.o.d -o src/CMakeFiles/llama.dir/models/olmo2.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/olmo2.cpp
[251/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/olmoe.cpp.o -MF src/CMakeFiles/llama.dir/models/olmoe.cpp.o.d -o src/CMakeFiles/llama.dir/models/olmoe.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/olmoe.cpp
[252/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/openai-moe-iswa.cpp.o -MF src/CMakeFiles/llama.dir/models/openai-moe-iswa.cpp.o.d -o src/CMakeFiles/llama.dir/models/openai-moe-iswa.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/openai-moe-iswa.cpp
[253/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/openelm.cpp.o -MF src/CMakeFiles/llama.dir/models/openelm.cpp.o.d -o src/CMakeFiles/llama.dir/models/openelm.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/openelm.cpp
[254/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/orion.cpp.o -MF src/CMakeFiles/llama.dir/models/orion.cpp.o.d -o src/CMakeFiles/llama.dir/models/orion.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/orion.cpp
[255/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/pangu-embedded.cpp.o -MF src/CMakeFiles/llama.dir/models/pangu-embedded.cpp.o.d -o src/CMakeFiles/llama.dir/models/pangu-embedded.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/pangu-embedded.cpp
[256/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/phi2.cpp.o -MF src/CMakeFiles/llama.dir/models/phi2.cpp.o.d -o src/CMakeFiles/llama.dir/models/phi2.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/phi2.cpp
[257/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/phi3.cpp.o -MF src/CMakeFiles/llama.dir/models/phi3.cpp.o.d -o src/CMakeFiles/llama.dir/models/phi3.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/phi3.cpp
[258/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/plamo.cpp.o -MF src/CMakeFiles/llama.dir/models/plamo.cpp.o.d -o src/CMakeFiles/llama.dir/models/plamo.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/plamo.cpp
[259/401] : && /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  -shared -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 -Xlinker --dependency-file=ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/link.d -Wl,-soname,libggml-hip.so.0 -o bin/libggml-hip.so.0.9.5 ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/acc.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/add-id.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/arange.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/argmax.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/argsort.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/binbcast.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/clamp.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/concat.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv-transpose-1d.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv2d-dw.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv2d-transpose.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/conv2d.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/convert.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/count-equal.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/cpy.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/cross-entropy-loss.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/cumsum.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/diag.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/diagmask.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-tile.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn-wmma-f16.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fattn.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/fill.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/getrows.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/ggml-cuda.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/gla.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/im2col.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mean.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmf.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmid.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmq.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmvf.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/mmvq.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/norm.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/opt-step-adamw.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/opt-step-sgd.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/out-prod.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/pad.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/pad_reflect_1d.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/pool2d.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/quantize.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/roll.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/rope.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/scale.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/set-rows.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/set.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/softcap.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/softmax.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/solve_tri.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/ssm-conv.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/ssm-scan.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/sum.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/sumrows.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/top-k.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/topk-moe.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/tri.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/tsembd.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/unary.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/upscale.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/wkv.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_16.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_32.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_1-ncols2_8.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_1.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_2.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_16-ncols2_4.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_16.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_32.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_4.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_2-ncols2_8.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_1.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_32-ncols2_2.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_16.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_2.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_4.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_4-ncols2_8.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_64-ncols2_1.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_1.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq1_s.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_s.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq2_xxs.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_s.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-mxfp4.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q2_k.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q3_k.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_0.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_1.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q4_k.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_0.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_1.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q5_k.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q6_k.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmq-instance-q8_0.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_1.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_10.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_11.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_12.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_13.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_14.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_15.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_16.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_2.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_3.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_4.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_5.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_6.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_7.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_8.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/mmf-instance-ncols_9.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-instance-q4_0-q4_0.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-instance-q8_0-q8_0.cu.o ggml/src/ggml-hip/CMakeFiles/ggml-hip.dir/__/ggml-cuda/template-instances/fattn-vec-instance-f16-f16.cu.o  bin/libggml-base.so.0.9.5  /usr/lib64/libhipblas.so.3.2  --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=gfx950  --offload-arch=gfx1010  --offload-arch=gfx1012  --offload-arch=gfx1030  --offload-arch=gfx1031  --offload-arch=gfx1035  --offload-arch=gfx1036  --offload-arch=gfx1100  --offload-arch=gfx1101  --offload-arch=gfx1102  --offload-arch=gfx1103  --offload-arch=gfx1150  --offload-arch=gfx1151  --offload-arch=gfx1152  --offload-arch=gfx1153  --offload-arch=gfx1200  --offload-arch=gfx1201  /usr/lib64/librocblas.so.5.2  /usr/lib64/rocm/llvm/lib/clang/22/lib/linux/libclang_rt.builtins-x86_64.a  /usr/lib64/libamdhip64.so.7.2.26095 && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
[260/401] /usr/bin/cmake -E cmake_symlink_library bin/libggml-hip.so.0.9.5  bin/libggml-hip.so.0 bin/libggml-hip.so && :
[261/401] : && /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  -shared -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 -Xlinker --dependency-file=ggml/src/CMakeFiles/ggml.dir/link.d -Wl,-soname,libggml.so.0 -o bin/libggml.so.0.9.5 ggml/src/CMakeFiles/ggml.dir/ggml-backend-dl.cpp.o ggml/src/CMakeFiles/ggml.dir/ggml-backend-reg.cpp.o  -ldl  bin/libggml-cpu.so.0.9.5  bin/libggml-hip.so.0.9.5  bin/libggml-base.so.0.9.5 && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
[262/401] /usr/bin/cmake -E cmake_symlink_library bin/libggml.so.0.9.5  bin/libggml.so.0 bin/libggml.so && :
[263/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/plamo2.cpp.o -MF src/CMakeFiles/llama.dir/models/plamo2.cpp.o.d -o src/CMakeFiles/llama.dir/models/plamo2.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/plamo2.cpp
[264/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/plamo3.cpp.o -MF src/CMakeFiles/llama.dir/models/plamo3.cpp.o.d -o src/CMakeFiles/llama.dir/models/plamo3.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/plamo3.cpp
[265/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/plm.cpp.o -MF src/CMakeFiles/llama.dir/models/plm.cpp.o.d -o src/CMakeFiles/llama.dir/models/plm.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/plm.cpp
[266/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/qwen.cpp.o -MF src/CMakeFiles/llama.dir/models/qwen.cpp.o.d -o src/CMakeFiles/llama.dir/models/qwen.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/qwen.cpp
[267/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/qwen2.cpp.o -MF src/CMakeFiles/llama.dir/models/qwen2.cpp.o.d -o src/CMakeFiles/llama.dir/models/qwen2.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/qwen2.cpp
[268/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/qwen2moe.cpp.o -MF src/CMakeFiles/llama.dir/models/qwen2moe.cpp.o.d -o src/CMakeFiles/llama.dir/models/qwen2moe.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/qwen2moe.cpp
[269/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/qwen2vl.cpp.o -MF src/CMakeFiles/llama.dir/models/qwen2vl.cpp.o.d -o src/CMakeFiles/llama.dir/models/qwen2vl.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/qwen2vl.cpp
[270/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/qwen3.cpp.o -MF src/CMakeFiles/llama.dir/models/qwen3.cpp.o.d -o src/CMakeFiles/llama.dir/models/qwen3.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/qwen3.cpp
[271/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/qwen3vl.cpp.o -MF src/CMakeFiles/llama.dir/models/qwen3vl.cpp.o.d -o src/CMakeFiles/llama.dir/models/qwen3vl.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/qwen3vl.cpp
[272/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/qwen3vl-moe.cpp.o -MF src/CMakeFiles/llama.dir/models/qwen3vl-moe.cpp.o.d -o src/CMakeFiles/llama.dir/models/qwen3vl-moe.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/qwen3vl-moe.cpp
[273/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/qwen3moe.cpp.o -MF src/CMakeFiles/llama.dir/models/qwen3moe.cpp.o.d -o src/CMakeFiles/llama.dir/models/qwen3moe.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/qwen3moe.cpp
[274/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/qwen3next.cpp.o -MF src/CMakeFiles/llama.dir/models/qwen3next.cpp.o.d -o src/CMakeFiles/llama.dir/models/qwen3next.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/qwen3next.cpp
[275/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/qwen35.cpp.o -MF src/CMakeFiles/llama.dir/models/qwen35.cpp.o.d -o src/CMakeFiles/llama.dir/models/qwen35.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/qwen35.cpp
[276/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/qwen35moe.cpp.o -MF src/CMakeFiles/llama.dir/models/qwen35moe.cpp.o.d -o src/CMakeFiles/llama.dir/models/qwen35moe.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/qwen35moe.cpp
[277/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/refact.cpp.o -MF src/CMakeFiles/llama.dir/models/refact.cpp.o.d -o src/CMakeFiles/llama.dir/models/refact.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/refact.cpp
[278/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/rnd1.cpp.o -MF src/CMakeFiles/llama.dir/models/rnd1.cpp.o.d -o src/CMakeFiles/llama.dir/models/rnd1.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/rnd1.cpp
[279/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/rwkv6.cpp.o -MF src/CMakeFiles/llama.dir/models/rwkv6.cpp.o.d -o src/CMakeFiles/llama.dir/models/rwkv6.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/rwkv6.cpp
[280/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/rwkv6-base.cpp.o -MF src/CMakeFiles/llama.dir/models/rwkv6-base.cpp.o.d -o src/CMakeFiles/llama.dir/models/rwkv6-base.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/rwkv6-base.cpp
[281/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/rwkv6qwen2.cpp.o -MF src/CMakeFiles/llama.dir/models/rwkv6qwen2.cpp.o.d -o src/CMakeFiles/llama.dir/models/rwkv6qwen2.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/rwkv6qwen2.cpp
[282/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/rwkv7-base.cpp.o -MF src/CMakeFiles/llama.dir/models/rwkv7-base.cpp.o.d -o src/CMakeFiles/llama.dir/models/rwkv7-base.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/rwkv7-base.cpp
[283/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/rwkv7.cpp.o -MF src/CMakeFiles/llama.dir/models/rwkv7.cpp.o.d -o src/CMakeFiles/llama.dir/models/rwkv7.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/rwkv7.cpp
[284/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/seed-oss.cpp.o -MF src/CMakeFiles/llama.dir/models/seed-oss.cpp.o.d -o src/CMakeFiles/llama.dir/models/seed-oss.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/seed-oss.cpp
[285/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/smallthinker.cpp.o -MF src/CMakeFiles/llama.dir/models/smallthinker.cpp.o.d -o src/CMakeFiles/llama.dir/models/smallthinker.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/smallthinker.cpp
[286/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/smollm3.cpp.o -MF src/CMakeFiles/llama.dir/models/smollm3.cpp.o.d -o src/CMakeFiles/llama.dir/models/smollm3.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/smollm3.cpp
[287/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/stablelm.cpp.o -MF src/CMakeFiles/llama.dir/models/stablelm.cpp.o.d -o src/CMakeFiles/llama.dir/models/stablelm.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/stablelm.cpp
[288/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/starcoder.cpp.o -MF src/CMakeFiles/llama.dir/models/starcoder.cpp.o.d -o src/CMakeFiles/llama.dir/models/starcoder.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/starcoder.cpp
[289/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/starcoder2.cpp.o -MF src/CMakeFiles/llama.dir/models/starcoder2.cpp.o.d -o src/CMakeFiles/llama.dir/models/starcoder2.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/starcoder2.cpp
[290/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/step35-iswa.cpp.o -MF src/CMakeFiles/llama.dir/models/step35-iswa.cpp.o.d -o src/CMakeFiles/llama.dir/models/step35-iswa.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/step35-iswa.cpp
[291/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/t5-dec.cpp.o -MF src/CMakeFiles/llama.dir/models/t5-dec.cpp.o.d -o src/CMakeFiles/llama.dir/models/t5-dec.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/t5-dec.cpp
[292/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/t5-enc.cpp.o -MF src/CMakeFiles/llama.dir/models/t5-enc.cpp.o.d -o src/CMakeFiles/llama.dir/models/t5-enc.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/t5-enc.cpp
[293/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/wavtokenizer-dec.cpp.o -MF src/CMakeFiles/llama.dir/models/wavtokenizer-dec.cpp.o.d -o src/CMakeFiles/llama.dir/models/wavtokenizer-dec.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/wavtokenizer-dec.cpp
[294/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/xverse.cpp.o -MF src/CMakeFiles/llama.dir/models/xverse.cpp.o.d -o src/CMakeFiles/llama.dir/models/xverse.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/xverse.cpp
[295/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/mistral3.cpp.o -MF src/CMakeFiles/llama.dir/models/mistral3.cpp.o.d -o src/CMakeFiles/llama.dir/models/mistral3.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/mistral3.cpp
[296/401] /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-b8064-build/llama.cpp-b8064/src/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/models/graph-context-mamba.cpp.o -MF src/CMakeFiles/llama.dir/models/graph-context-mamba.cpp.o.d -o src/CMakeFiles/llama.dir/models/graph-context-mamba.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/models/graph-context-mamba.cpp
[297/401] /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dmtmd_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wno-cast-qual -MD -MT tools/mtmd/CMakeFiles/mtmd.dir/mtmd.cpp.o -MF tools/mtmd/CMakeFiles/mtmd.dir/mtmd.cpp.o.d -o tools/mtmd/CMakeFiles/mtmd.dir/mtmd.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/mtmd.cpp
[298/401] /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dmtmd_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wno-cast-qual -MD -MT tools/mtmd/CMakeFiles/mtmd.dir/mtmd-audio.cpp.o -MF tools/mtmd/CMakeFiles/mtmd.dir/mtmd-audio.cpp.o.d -o tools/mtmd/CMakeFiles/mtmd.dir/mtmd-audio.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/mtmd-audio.cpp
[299/401] /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dmtmd_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wno-cast-qual -MD -MT tools/mtmd/CMakeFiles/mtmd.dir/clip.cpp.o -MF tools/mtmd/CMakeFiles/mtmd.dir/clip.cpp.o.d -o tools/mtmd/CMakeFiles/mtmd.dir/clip.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/clip.cpp
[300/401] /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dmtmd_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wno-cast-qual -MD -MT tools/mtmd/CMakeFiles/mtmd.dir/models/cogvlm.cpp.o -MF tools/mtmd/CMakeFiles/mtmd.dir/models/cogvlm.cpp.o.d -o tools/mtmd/CMakeFiles/mtmd.dir/models/cogvlm.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/models/cogvlm.cpp
[301/401] /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dmtmd_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wno-cast-qual -MD -MT tools/mtmd/CMakeFiles/mtmd.dir/mtmd-helper.cpp.o -MF tools/mtmd/CMakeFiles/mtmd.dir/mtmd-helper.cpp.o.d -o tools/mtmd/CMakeFiles/mtmd.dir/mtmd-helper.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/mtmd-helper.cpp
[302/401] /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dmtmd_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wno-cast-qual -MD -MT tools/mtmd/CMakeFiles/mtmd.dir/models/conformer.cpp.o -MF tools/mtmd/CMakeFiles/mtmd.dir/models/conformer.cpp.o.d -o tools/mtmd/CMakeFiles/mtmd.dir/models/conformer.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/models/conformer.cpp
[303/401] /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dmtmd_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wno-cast-qual -MD -MT tools/mtmd/CMakeFiles/mtmd.dir/models/glm4v.cpp.o -MF tools/mtmd/CMakeFiles/mtmd.dir/models/glm4v.cpp.o.d -o tools/mtmd/CMakeFiles/mtmd.dir/models/glm4v.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/models/glm4v.cpp
[304/401] /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dmtmd_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wno-cast-qual -MD -MT tools/mtmd/CMakeFiles/mtmd.dir/models/internvl.cpp.o -MF tools/mtmd/CMakeFiles/mtmd.dir/models/internvl.cpp.o.d -o tools/mtmd/CMakeFiles/mtmd.dir/models/internvl.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/models/internvl.cpp
[305/401] /usr/bin/hipcc -DCPPHTTPLIB_FORM_URL_ENCODED_PAYLOAD_MAX_LENGTH=1048576 -DCPPHTTPLIB_LISTEN_BACKLOG=512 -DCPPHTTPLIB_OPENSSL_SUPPORT -DCPPHTTPLIB_REQUEST_URI_MAX_LENGTH=32768 -DCPPHTTPLIB_TCP_NODELAY=1  -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 -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -w -MD -MT vendor/cpp-httplib/CMakeFiles/cpp-httplib.dir/httplib.cpp.o -MF vendor/cpp-httplib/CMakeFiles/cpp-httplib.dir/httplib.cpp.o.d -o vendor/cpp-httplib/CMakeFiles/cpp-httplib.dir/httplib.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/vendor/cpp-httplib/httplib.cpp
[306/401] /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dmtmd_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wno-cast-qual -MD -MT tools/mtmd/CMakeFiles/mtmd.dir/models/kimivl.cpp.o -MF tools/mtmd/CMakeFiles/mtmd.dir/models/kimivl.cpp.o.d -o tools/mtmd/CMakeFiles/mtmd.dir/models/kimivl.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/models/kimivl.cpp
[307/401] /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dmtmd_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wno-cast-qual -MD -MT tools/mtmd/CMakeFiles/mtmd.dir/models/kimik25.cpp.o -MF tools/mtmd/CMakeFiles/mtmd.dir/models/kimik25.cpp.o.d -o tools/mtmd/CMakeFiles/mtmd.dir/models/kimik25.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/models/kimik25.cpp
[308/401] /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dmtmd_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wno-cast-qual -MD -MT tools/mtmd/CMakeFiles/mtmd.dir/models/nemotron-v2-vl.cpp.o -MF tools/mtmd/CMakeFiles/mtmd.dir/models/nemotron-v2-vl.cpp.o.d -o tools/mtmd/CMakeFiles/mtmd.dir/models/nemotron-v2-vl.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/models/nemotron-v2-vl.cpp
[309/401] /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dmtmd_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wno-cast-qual -MD -MT tools/mtmd/CMakeFiles/mtmd.dir/models/llama4.cpp.o -MF tools/mtmd/CMakeFiles/mtmd.dir/models/llama4.cpp.o.d -o tools/mtmd/CMakeFiles/mtmd.dir/models/llama4.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/models/llama4.cpp
[310/401] /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dmtmd_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wno-cast-qual -MD -MT tools/mtmd/CMakeFiles/mtmd.dir/models/minicpmv.cpp.o -MF tools/mtmd/CMakeFiles/mtmd.dir/models/minicpmv.cpp.o.d -o tools/mtmd/CMakeFiles/mtmd.dir/models/minicpmv.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/models/minicpmv.cpp
[311/401] /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dmtmd_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wno-cast-qual -MD -MT tools/mtmd/CMakeFiles/mtmd.dir/models/llava.cpp.o -MF tools/mtmd/CMakeFiles/mtmd.dir/models/llava.cpp.o.d -o tools/mtmd/CMakeFiles/mtmd.dir/models/llava.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/models/llava.cpp
[312/401] /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dmtmd_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wno-cast-qual -MD -MT tools/mtmd/CMakeFiles/mtmd.dir/models/pixtral.cpp.o -MF tools/mtmd/CMakeFiles/mtmd.dir/models/pixtral.cpp.o.d -o tools/mtmd/CMakeFiles/mtmd.dir/models/pixtral.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/models/pixtral.cpp
[313/401] /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dmtmd_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wno-cast-qual -MD -MT tools/mtmd/CMakeFiles/mtmd.dir/models/qwen2vl.cpp.o -MF tools/mtmd/CMakeFiles/mtmd.dir/models/qwen2vl.cpp.o.d -o tools/mtmd/CMakeFiles/mtmd.dir/models/qwen2vl.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/models/qwen2vl.cpp
[314/401] /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dmtmd_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wno-cast-qual -MD -MT tools/mtmd/CMakeFiles/mtmd.dir/models/qwen3vl.cpp.o -MF tools/mtmd/CMakeFiles/mtmd.dir/models/qwen3vl.cpp.o.d -o tools/mtmd/CMakeFiles/mtmd.dir/models/qwen3vl.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/models/qwen3vl.cpp
[315/401] /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dmtmd_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wno-cast-qual -MD -MT tools/mtmd/CMakeFiles/mtmd.dir/models/siglip.cpp.o -MF tools/mtmd/CMakeFiles/mtmd.dir/models/siglip.cpp.o.d -o tools/mtmd/CMakeFiles/mtmd.dir/models/siglip.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/models/siglip.cpp
[316/401] /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dmtmd_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wno-cast-qual -MD -MT tools/mtmd/CMakeFiles/mtmd.dir/models/whisper-enc.cpp.o -MF tools/mtmd/CMakeFiles/mtmd.dir/models/whisper-enc.cpp.o.d -o tools/mtmd/CMakeFiles/mtmd.dir/models/whisper-enc.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/models/whisper-enc.cpp
[317/401] /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dmtmd_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wno-cast-qual -MD -MT tools/mtmd/CMakeFiles/mtmd.dir/models/mobilenetv5.cpp.o -MF tools/mtmd/CMakeFiles/mtmd.dir/models/mobilenetv5.cpp.o.d -o tools/mtmd/CMakeFiles/mtmd.dir/models/mobilenetv5.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/models/mobilenetv5.cpp
[318/401] /usr/bin/hipcc -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_BUILD -DLLAMA_SHARED -Dmtmd_EXPORTS -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../.. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/../../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wno-cast-qual -MD -MT tools/mtmd/CMakeFiles/mtmd.dir/models/youtuvl.cpp.o -MF tools/mtmd/CMakeFiles/mtmd.dir/models/youtuvl.cpp.o.d -o tools/mtmd/CMakeFiles/mtmd.dir/models/youtuvl.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/models/youtuvl.cpp
[319/401] /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 common/CMakeFiles/build_info.dir/build-info.cpp.o.d -o common/CMakeFiles/build_info.dir/build-info.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/redhat-linux-build/common/build-info.cpp
[320/401] : && /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  -shared -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 -Xlinker --dependency-file=src/CMakeFiles/llama.dir/link.d -Wl,-soname,libllama.so.0 -o bin/libllama.so.0.0.0 src/CMakeFiles/llama.dir/llama.cpp.o src/CMakeFiles/llama.dir/llama-adapter.cpp.o src/CMakeFiles/llama.dir/llama-arch.cpp.o src/CMakeFiles/llama.dir/llama-batch.cpp.o src/CMakeFiles/llama.dir/llama-chat.cpp.o src/CMakeFiles/llama.dir/llama-context.cpp.o src/CMakeFiles/llama.dir/llama-cparams.cpp.o src/CMakeFiles/llama.dir/llama-grammar.cpp.o src/CMakeFiles/llama.dir/llama-graph.cpp.o src/CMakeFiles/llama.dir/llama-hparams.cpp.o src/CMakeFiles/llama.dir/llama-impl.cpp.o src/CMakeFiles/llama.dir/llama-io.cpp.o src/CMakeFiles/llama.dir/llama-kv-cache.cpp.o src/CMakeFiles/llama.dir/llama-kv-cache-iswa.cpp.o src/CMakeFiles/llama.dir/llama-memory.cpp.o src/CMakeFiles/llama.dir/llama-memory-hybrid.cpp.o src/CMakeFiles/llama.dir/llama-memory-hybrid-iswa.cpp.o src/CMakeFiles/llama.dir/llama-memory-recurrent.cpp.o src/CMakeFiles/llama.dir/llama-mmap.cpp.o src/CMakeFiles/llama.dir/llama-model-loader.cpp.o src/CMakeFiles/llama.dir/llama-model-saver.cpp.o src/CMakeFiles/llama.dir/llama-model.cpp.o src/CMakeFiles/llama.dir/llama-quant.cpp.o src/CMakeFiles/llama.dir/llama-sampler.cpp.o src/CMakeFiles/llama.dir/llama-vocab.cpp.o src/CMakeFiles/llama.dir/unicode-data.cpp.o src/CMakeFiles/llama.dir/unicode.cpp.o src/CMakeFiles/llama.dir/models/afmoe.cpp.o src/CMakeFiles/llama.dir/models/apertus.cpp.o src/CMakeFiles/llama.dir/models/arcee.cpp.o src/CMakeFiles/llama.dir/models/arctic.cpp.o src/CMakeFiles/llama.dir/models/arwkv7.cpp.o src/CMakeFiles/llama.dir/models/baichuan.cpp.o src/CMakeFiles/llama.dir/models/bailingmoe.cpp.o src/CMakeFiles/llama.dir/models/bailingmoe2.cpp.o src/CMakeFiles/llama.dir/models/bert.cpp.o src/CMakeFiles/llama.dir/models/bitnet.cpp.o src/CMakeFiles/llama.dir/models/bloom.cpp.o src/CMakeFiles/llama.dir/models/chameleon.cpp.o src/CMakeFiles/llama.dir/models/chatglm.cpp.o src/CMakeFiles/llama.dir/models/codeshell.cpp.o src/CMakeFiles/llama.dir/models/cogvlm.cpp.o src/CMakeFiles/llama.dir/models/cohere2-iswa.cpp.o src/CMakeFiles/llama.dir/models/command-r.cpp.o src/CMakeFiles/llama.dir/models/dbrx.cpp.o src/CMakeFiles/llama.dir/models/deci.cpp.o src/CMakeFiles/llama.dir/models/deepseek.cpp.o src/CMakeFiles/llama.dir/models/deepseek2.cpp.o src/CMakeFiles/llama.dir/models/dots1.cpp.o src/CMakeFiles/llama.dir/models/dream.cpp.o src/CMakeFiles/llama.dir/models/ernie4-5-moe.cpp.o src/CMakeFiles/llama.dir/models/ernie4-5.cpp.o src/CMakeFiles/llama.dir/models/exaone.cpp.o src/CMakeFiles/llama.dir/models/exaone4.cpp.o src/CMakeFiles/llama.dir/models/exaone-moe.cpp.o src/CMakeFiles/llama.dir/models/falcon-h1.cpp.o src/CMakeFiles/llama.dir/models/falcon.cpp.o src/CMakeFiles/llama.dir/models/gemma-embedding.cpp.o src/CMakeFiles/llama.dir/models/gemma.cpp.o src/CMakeFiles/llama.dir/models/gemma2-iswa.cpp.o src/CMakeFiles/llama.dir/models/gemma3.cpp.o src/CMakeFiles/llama.dir/models/gemma3n-iswa.cpp.o src/CMakeFiles/llama.dir/models/glm4-moe.cpp.o src/CMakeFiles/llama.dir/models/glm4.cpp.o src/CMakeFiles/llama.dir/models/gpt2.cpp.o src/CMakeFiles/llama.dir/models/gptneox.cpp.o src/CMakeFiles/llama.dir/models/granite-hybrid.cpp.o src/CMakeFiles/llama.dir/models/granite.cpp.o src/CMakeFiles/llama.dir/models/grok.cpp.o src/CMakeFiles/llama.dir/models/grovemoe.cpp.o src/CMakeFiles/llama.dir/models/hunyuan-dense.cpp.o src/CMakeFiles/llama.dir/models/hunyuan-moe.cpp.o src/CMakeFiles/llama.dir/models/internlm2.cpp.o src/CMakeFiles/llama.dir/models/jais.cpp.o src/CMakeFiles/llama.dir/models/jamba.cpp.o src/CMakeFiles/llama.dir/models/kimi-linear.cpp.o src/CMakeFiles/llama.dir/models/lfm2.cpp.o src/CMakeFiles/llama.dir/models/llada-moe.cpp.o src/CMakeFiles/llama.dir/models/llada.cpp.o src/CMakeFiles/llama.dir/models/llama-iswa.cpp.o src/CMakeFiles/llama.dir/models/llama.cpp.o src/CMakeFiles/llama.dir/models/maincoder.cpp.o src/CMakeFiles/llama.dir/models/mamba.cpp.o src/CMakeFiles/llama.dir/models/mimo2-iswa.cpp.o src/CMakeFiles/llama.dir/models/minicpm3.cpp.o src/CMakeFiles/llama.dir/models/minimax-m2.cpp.o src/CMakeFiles/llama.dir/models/modern-bert.cpp.o src/CMakeFiles/llama.dir/models/mpt.cpp.o src/CMakeFiles/llama.dir/models/nemotron-h.cpp.o src/CMakeFiles/llama.dir/models/nemotron.cpp.o src/CMakeFiles/llama.dir/models/neo-bert.cpp.o src/CMakeFiles/llama.dir/models/olmo.cpp.o src/CMakeFiles/llama.dir/models/olmo2.cpp.o src/CMakeFiles/llama.dir/models/olmoe.cpp.o src/CMakeFiles/llama.dir/models/openai-moe-iswa.cpp.o src/CMakeFiles/llama.dir/models/openelm.cpp.o src/CMakeFiles/llama.dir/models/orion.cpp.o src/CMakeFiles/llama.dir/models/pangu-embedded.cpp.o src/CMakeFiles/llama.dir/models/phi2.cpp.o src/CMakeFiles/llama.dir/models/phi3.cpp.o src/CMakeFiles/llama.dir/models/plamo.cpp.o src/CMakeFiles/llama.dir/models/plamo2.cpp.o src/CMakeFiles/llama.dir/models/plamo3.cpp.o src/CMakeFiles/llama.dir/models/plm.cpp.o src/CMakeFiles/llama.dir/models/qwen.cpp.o src/CMakeFiles/llama.dir/models/qwen2.cpp.o src/CMakeFiles/llama.dir/models/qwen2moe.cpp.o src/CMakeFiles/llama.dir/models/qwen2vl.cpp.o src/CMakeFiles/llama.dir/models/qwen3.cpp.o src/CMakeFiles/llama.dir/models/qwen3vl.cpp.o src/CMakeFiles/llama.dir/models/qwen3vl-moe.cpp.o src/CMakeFiles/llama.dir/models/qwen3moe.cpp.o src/CMakeFiles/llama.dir/models/qwen3next.cpp.o src/CMakeFiles/llama.dir/models/qwen35.cpp.o src/CMakeFiles/llama.dir/models/qwen35moe.cpp.o src/CMakeFiles/llama.dir/models/refact.cpp.o src/CMakeFiles/llama.dir/models/rnd1.cpp.o src/CMakeFiles/llama.dir/models/rwkv6-base.cpp.o src/CMakeFiles/llama.dir/models/rwkv6.cpp.o src/CMakeFiles/llama.dir/models/rwkv6qwen2.cpp.o src/CMakeFiles/llama.dir/models/rwkv7-base.cpp.o src/CMakeFiles/llama.dir/models/rwkv7.cpp.o src/CMakeFiles/llama.dir/models/seed-oss.cpp.o src/CMakeFiles/llama.dir/models/smallthinker.cpp.o src/CMakeFiles/llama.dir/models/smollm3.cpp.o src/CMakeFiles/llama.dir/models/stablelm.cpp.o src/CMakeFiles/llama.dir/models/starcoder.cpp.o src/CMakeFiles/llama.dir/models/starcoder2.cpp.o src/CMakeFiles/llama.dir/models/step35-iswa.cpp.o src/CMakeFiles/llama.dir/models/t5-dec.cpp.o src/CMakeFiles/llama.dir/models/t5-enc.cpp.o src/CMakeFiles/llama.dir/models/wavtokenizer-dec.cpp.o src/CMakeFiles/llama.dir/models/xverse.cpp.o src/CMakeFiles/llama.dir/models/mistral3.cpp.o src/CMakeFiles/llama.dir/models/graph-context-mamba.cpp.o  bin/libggml.so.0.9.5  bin/libggml-cpu.so.0.9.5  bin/libggml-hip.so.0.9.5  bin/libggml-base.so.0.9.5 && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
[321/401] /usr/bin/cmake -E cmake_symlink_library bin/libllama.so.0.0.0  bin/libllama.so.0 bin/libllama.so && :
[322/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/chat-parser.cpp.o -MF common/CMakeFiles/common.dir/chat-parser.cpp.o.d -o common/CMakeFiles/common.dir/chat-parser.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/chat-parser.cpp
[323/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/chat-peg-parser.cpp.o -MF common/CMakeFiles/common.dir/chat-peg-parser.cpp.o.d -o common/CMakeFiles/common.dir/chat-peg-parser.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/chat-peg-parser.cpp
[324/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/chat-parser-xml-toolcall.cpp.o -MF common/CMakeFiles/common.dir/chat-parser-xml-toolcall.cpp.o.d -o common/CMakeFiles/common.dir/chat-parser-xml-toolcall.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/chat-parser-xml-toolcall.cpp
[325/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 common/CMakeFiles/common.dir/console.cpp.o.d -o common/CMakeFiles/common.dir/console.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/console.cpp
[326/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 common/CMakeFiles/common.dir/arg.cpp.o.d -o common/CMakeFiles/common.dir/arg.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/arg.cpp
[327/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 common/CMakeFiles/common.dir/common.cpp.o.d -o common/CMakeFiles/common.dir/common.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/common.cpp
[328/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/debug.cpp.o -MF common/CMakeFiles/common.dir/debug.cpp.o.d -o common/CMakeFiles/common.dir/debug.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/debug.cpp
[329/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-partial.cpp.o -MF common/CMakeFiles/common.dir/json-partial.cpp.o.d -o common/CMakeFiles/common.dir/json-partial.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/json-partial.cpp
[330/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/download.cpp.o -MF common/CMakeFiles/common.dir/download.cpp.o.d -o common/CMakeFiles/common.dir/download.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/download.cpp
[331/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/llguidance.cpp.o -MF common/CMakeFiles/common.dir/llguidance.cpp.o.d -o common/CMakeFiles/common.dir/llguidance.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/llguidance.cpp
[332/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 common/CMakeFiles/common.dir/log.cpp.o.d -o common/CMakeFiles/common.dir/log.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/log.cpp
[333/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 common/CMakeFiles/common.dir/ngram-cache.cpp.o.d -o common/CMakeFiles/common.dir/ngram-cache.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/ngram-cache.cpp
[334/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 common/CMakeFiles/common.dir/json-schema-to-grammar.cpp.o.d -o common/CMakeFiles/common.dir/json-schema-to-grammar.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/json-schema-to-grammar.cpp
[335/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-map.cpp.o -MF common/CMakeFiles/common.dir/ngram-map.cpp.o.d -o common/CMakeFiles/common.dir/ngram-map.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/ngram-map.cpp
[336/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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-mod.cpp.o -MF common/CMakeFiles/common.dir/ngram-mod.cpp.o.d -o common/CMakeFiles/common.dir/ngram-mod.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/ngram-mod.cpp
[337/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/chat.cpp.o -MF common/CMakeFiles/common.dir/chat.cpp.o.d -o common/CMakeFiles/common.dir/chat.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/chat.cpp
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/chat.cpp:10:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/parser.h:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:130:21: warning: function 'as_int' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  130 |     virtual int64_t as_int() const { throw std::runtime_error(type() + " is not an int value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:131:20: warning: function 'as_float' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  131 |     virtual double as_float() const { throw std::runtime_error(type() + " is not a float value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:132:20: warning: function 'as_string' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  132 |     virtual string as_string() const { throw std::runtime_error(type() + " is not a string value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:133:18: warning: function 'as_bool' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  133 |     virtual bool as_bool() const { throw std::runtime_error(type() + " is not a bool value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:134:40: warning: function 'as_array' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  134 |     virtual const std::vector<value> & as_array() const { throw std::runtime_error(type() + " is not an array value"); }
      |                                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:135:58: warning: function 'as_ordered_object' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  135 |     virtual const std::vector<std::pair<value, value>> & as_ordered_object() const { throw std::runtime_error(type() + " is not an object value"); }
      |                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:136:19: warning: function 'invoke' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  136 |     virtual value invoke(const func_args &) const { throw std::runtime_error(type() + " is not a function value"); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:139:35: warning: function 'get_builtins' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  139 |     virtual const func_builtins & get_builtins() const {
      |                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:143:18: warning: function 'has_key' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  143 |     virtual bool has_key(const value &) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:144:18: warning: function 'insert' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  144 |     virtual void insert(const value & /* key */, const value & /* val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:145:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  145 |     virtual value & at(const value & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:146:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     virtual value & at(const value & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:147:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  147 |     virtual value & at(const std::string & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:148:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  148 |     virtual value & at(const std::string & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:149:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  149 |     virtual value & at(int64_t /* idx */, value & /* default_val */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:150:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  150 |     virtual value & at(int64_t /* idx */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/chat.cpp:10:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/parser.h:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:110:19: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  110 |     virtual value execute_impl(context &) { throw std::runtime_error("cannot exec " + type()); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:146:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:198:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  198 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:212:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  212 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:504:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  504 |     value execute_impl(context &) override {
      |           ^
21 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/chat.cpp:10:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/parser.h:4:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:130:21: warning: function 'as_int' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  130 |     virtual int64_t as_int() const { throw std::runtime_error(type() + " is not an int value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:131:20: warning: function 'as_float' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  131 |     virtual double as_float() const { throw std::runtime_error(type() + " is not a float value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:132:20: warning: function 'as_string' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  132 |     virtual string as_string() const { throw std::runtime_error(type() + " is not a string value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:133:18: warning: function 'as_bool' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  133 |     virtual bool as_bool() const { throw std::runtime_error(type() + " is not a bool value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:134:40: warning: function 'as_array' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  134 |     virtual const std::vector<value> & as_array() const { throw std::runtime_error(type() + " is not an array value"); }
      |                                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:135:58: warning: function 'as_ordered_object' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  135 |     virtual const std::vector<std::pair<value, value>> & as_ordered_object() const { throw std::runtime_error(type() + " is not an object value"); }
      |                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:136:19: warning: function 'invoke' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  136 |     virtual value invoke(const func_args &) const { throw std::runtime_error(type() + " is not a function value"); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:139:35: warning: function 'get_builtins' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  139 |     virtual const func_builtins & get_builtins() const {
      |                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:143:18: warning: function 'has_key' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  143 |     virtual bool has_key(const value &) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:144:18: warning: function 'insert' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  144 |     virtual void insert(const value & /* key */, const value & /* val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:145:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  145 |     virtual value & at(const value & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:146:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     virtual value & at(const value & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:147:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  147 |     virtual value & at(const std::string & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:148:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  148 |     virtual value & at(const std::string & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:149:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  149 |     virtual value & at(int64_t /* idx */, value & /* default_val */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:150:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  150 |     virtual value & at(int64_t /* idx */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/chat.cpp:10:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/parser.h:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:110:19: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  110 |     virtual value execute_impl(context &) { throw std::runtime_error("cannot exec " + type()); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:146:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:198:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  198 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:212:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  212 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:504:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  504 |     value execute_impl(context &) override {
      |           ^
21 warnings generated when compiling for host.
[338/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 common/CMakeFiles/common.dir/sampling.cpp.o.d -o common/CMakeFiles/common.dir/sampling.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/sampling.cpp
[339/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/preset.cpp.o -MF common/CMakeFiles/common.dir/preset.cpp.o.d -o common/CMakeFiles/common.dir/preset.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/preset.cpp
[340/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/regex-partial.cpp.o -MF common/CMakeFiles/common.dir/regex-partial.cpp.o.d -o common/CMakeFiles/common.dir/regex-partial.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/regex-partial.cpp
[341/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/unicode.cpp.o -MF common/CMakeFiles/common.dir/unicode.cpp.o.d -o common/CMakeFiles/common.dir/unicode.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/unicode.cpp
[342/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 common/CMakeFiles/common.dir/speculative.cpp.o.d -o common/CMakeFiles/common.dir/speculative.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/speculative.cpp
[343/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/jinja/lexer.cpp.o -MF common/CMakeFiles/common.dir/jinja/lexer.cpp.o.d -o common/CMakeFiles/common.dir/jinja/lexer.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/lexer.cpp
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/lexer.cpp:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:130:21: warning: function 'as_int' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  130 |     virtual int64_t as_int() const { throw std::runtime_error(type() + " is not an int value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:131:20: warning: function 'as_float' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  131 |     virtual double as_float() const { throw std::runtime_error(type() + " is not a float value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:132:20: warning: function 'as_string' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  132 |     virtual string as_string() const { throw std::runtime_error(type() + " is not a string value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:133:18: warning: function 'as_bool' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  133 |     virtual bool as_bool() const { throw std::runtime_error(type() + " is not a bool value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:134:40: warning: function 'as_array' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  134 |     virtual const std::vector<value> & as_array() const { throw std::runtime_error(type() + " is not an array value"); }
      |                                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:135:58: warning: function 'as_ordered_object' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  135 |     virtual const std::vector<std::pair<value, value>> & as_ordered_object() const { throw std::runtime_error(type() + " is not an object value"); }
      |                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:136:19: warning: function 'invoke' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  136 |     virtual value invoke(const func_args &) const { throw std::runtime_error(type() + " is not a function value"); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:139:35: warning: function 'get_builtins' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  139 |     virtual const func_builtins & get_builtins() const {
      |                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:143:18: warning: function 'has_key' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  143 |     virtual bool has_key(const value &) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:144:18: warning: function 'insert' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  144 |     virtual void insert(const value & /* key */, const value & /* val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:145:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  145 |     virtual value & at(const value & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:146:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     virtual value & at(const value & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:147:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  147 |     virtual value & at(const std::string & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:148:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  148 |     virtual value & at(const std::string & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:149:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  149 |     virtual value & at(int64_t /* idx */, value & /* default_val */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:150:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  150 |     virtual value & at(int64_t /* idx */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/lexer.cpp:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:110:19: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  110 |     virtual value execute_impl(context &) { throw std::runtime_error("cannot exec " + type()); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:146:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:198:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  198 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:212:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  212 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:504:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  504 |     value execute_impl(context &) override {
      |           ^
21 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/lexer.cpp:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:130:21: warning: function 'as_int' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  130 |     virtual int64_t as_int() const { throw std::runtime_error(type() + " is not an int value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:131:20: warning: function 'as_float' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  131 |     virtual double as_float() const { throw std::runtime_error(type() + " is not a float value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:132:20: warning: function 'as_string' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  132 |     virtual string as_string() const { throw std::runtime_error(type() + " is not a string value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:133:18: warning: function 'as_bool' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  133 |     virtual bool as_bool() const { throw std::runtime_error(type() + " is not a bool value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:134:40: warning: function 'as_array' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  134 |     virtual const std::vector<value> & as_array() const { throw std::runtime_error(type() + " is not an array value"); }
      |                                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:135:58: warning: function 'as_ordered_object' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  135 |     virtual const std::vector<std::pair<value, value>> & as_ordered_object() const { throw std::runtime_error(type() + " is not an object value"); }
      |                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:136:19: warning: function 'invoke' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  136 |     virtual value invoke(const func_args &) const { throw std::runtime_error(type() + " is not a function value"); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:139:35: warning: function 'get_builtins' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  139 |     virtual const func_builtins & get_builtins() const {
      |                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:143:18: warning: function 'has_key' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  143 |     virtual bool has_key(const value &) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:144:18: warning: function 'insert' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  144 |     virtual void insert(const value & /* key */, const value & /* val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:145:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  145 |     virtual value & at(const value & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:146:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     virtual value & at(const value & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:147:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  147 |     virtual value & at(const std::string & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:148:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  148 |     virtual value & at(const std::string & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:149:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  149 |     virtual value & at(int64_t /* idx */, value & /* default_val */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:150:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  150 |     virtual value & at(int64_t /* idx */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/lexer.cpp:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:110:19: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  110 |     virtual value execute_impl(context &) { throw std::runtime_error("cannot exec " + type()); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:146:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:198:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  198 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:212:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  212 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:504:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  504 |     value execute_impl(context &) override {
      |           ^
21 warnings generated when compiling for host.
[344/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/jinja/parser.cpp.o -MF common/CMakeFiles/common.dir/jinja/parser.cpp.o.d -o common/CMakeFiles/common.dir/jinja/parser.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/parser.cpp
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/parser.cpp:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:130:21: warning: function 'as_int' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  130 |     virtual int64_t as_int() const { throw std::runtime_error(type() + " is not an int value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:131:20: warning: function 'as_float' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  131 |     virtual double as_float() const { throw std::runtime_error(type() + " is not a float value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:132:20: warning: function 'as_string' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  132 |     virtual string as_string() const { throw std::runtime_error(type() + " is not a string value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:133:18: warning: function 'as_bool' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  133 |     virtual bool as_bool() const { throw std::runtime_error(type() + " is not a bool value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:134:40: warning: function 'as_array' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  134 |     virtual const std::vector<value> & as_array() const { throw std::runtime_error(type() + " is not an array value"); }
      |                                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:135:58: warning: function 'as_ordered_object' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  135 |     virtual const std::vector<std::pair<value, value>> & as_ordered_object() const { throw std::runtime_error(type() + " is not an object value"); }
      |                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:136:19: warning: function 'invoke' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  136 |     virtual value invoke(const func_args &) const { throw std::runtime_error(type() + " is not a function value"); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:139:35: warning: function 'get_builtins' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  139 |     virtual const func_builtins & get_builtins() const {
      |                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:143:18: warning: function 'has_key' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  143 |     virtual bool has_key(const value &) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:144:18: warning: function 'insert' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  144 |     virtual void insert(const value & /* key */, const value & /* val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:145:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  145 |     virtual value & at(const value & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:146:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     virtual value & at(const value & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:147:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  147 |     virtual value & at(const std::string & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:148:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  148 |     virtual value & at(const std::string & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:149:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  149 |     virtual value & at(int64_t /* idx */, value & /* default_val */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:150:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  150 |     virtual value & at(int64_t /* idx */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/parser.cpp:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:110:19: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  110 |     virtual value execute_impl(context &) { throw std::runtime_error("cannot exec " + type()); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:146:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:198:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  198 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:212:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  212 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:504:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  504 |     value execute_impl(context &) override {
      |           ^
21 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/parser.cpp:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:130:21: warning: function 'as_int' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  130 |     virtual int64_t as_int() const { throw std::runtime_error(type() + " is not an int value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:131:20: warning: function 'as_float' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  131 |     virtual double as_float() const { throw std::runtime_error(type() + " is not a float value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:132:20: warning: function 'as_string' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  132 |     virtual string as_string() const { throw std::runtime_error(type() + " is not a string value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:133:18: warning: function 'as_bool' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  133 |     virtual bool as_bool() const { throw std::runtime_error(type() + " is not a bool value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:134:40: warning: function 'as_array' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  134 |     virtual const std::vector<value> & as_array() const { throw std::runtime_error(type() + " is not an array value"); }
      |                                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:135:58: warning: function 'as_ordered_object' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  135 |     virtual const std::vector<std::pair<value, value>> & as_ordered_object() const { throw std::runtime_error(type() + " is not an object value"); }
      |                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:136:19: warning: function 'invoke' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  136 |     virtual value invoke(const func_args &) const { throw std::runtime_error(type() + " is not a function value"); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:139:35: warning: function 'get_builtins' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  139 |     virtual const func_builtins & get_builtins() const {
      |                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:143:18: warning: function 'has_key' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  143 |     virtual bool has_key(const value &) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:144:18: warning: function 'insert' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  144 |     virtual void insert(const value & /* key */, const value & /* val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:145:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  145 |     virtual value & at(const value & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:146:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     virtual value & at(const value & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:147:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  147 |     virtual value & at(const std::string & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:148:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  148 |     virtual value & at(const std::string & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:149:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  149 |     virtual value & at(int64_t /* idx */, value & /* default_val */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:150:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  150 |     virtual value & at(int64_t /* idx */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/parser.cpp:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:110:19: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  110 |     virtual value execute_impl(context &) { throw std::runtime_error("cannot exec " + type()); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:146:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:198:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  198 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:212:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  212 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:504:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  504 |     value execute_impl(context &) override {
      |           ^
21 warnings generated when compiling for host.
[345/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/jinja/runtime.cpp.o -MF common/CMakeFiles/common.dir/jinja/runtime.cpp.o.d -o common/CMakeFiles/common.dir/jinja/runtime.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.cpp
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.cpp:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:130:21: warning: function 'as_int' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  130 |     virtual int64_t as_int() const { throw std::runtime_error(type() + " is not an int value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:131:20: warning: function 'as_float' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  131 |     virtual double as_float() const { throw std::runtime_error(type() + " is not a float value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:132:20: warning: function 'as_string' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  132 |     virtual string as_string() const { throw std::runtime_error(type() + " is not a string value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:133:18: warning: function 'as_bool' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  133 |     virtual bool as_bool() const { throw std::runtime_error(type() + " is not a bool value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:134:40: warning: function 'as_array' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  134 |     virtual const std::vector<value> & as_array() const { throw std::runtime_error(type() + " is not an array value"); }
      |                                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:135:58: warning: function 'as_ordered_object' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  135 |     virtual const std::vector<std::pair<value, value>> & as_ordered_object() const { throw std::runtime_error(type() + " is not an object value"); }
      |                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:136:19: warning: function 'invoke' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  136 |     virtual value invoke(const func_args &) const { throw std::runtime_error(type() + " is not a function value"); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:139:35: warning: function 'get_builtins' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  139 |     virtual const func_builtins & get_builtins() const {
      |                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:143:18: warning: function 'has_key' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  143 |     virtual bool has_key(const value &) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:144:18: warning: function 'insert' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  144 |     virtual void insert(const value & /* key */, const value & /* val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:145:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  145 |     virtual value & at(const value & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:146:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     virtual value & at(const value & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:147:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  147 |     virtual value & at(const std::string & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:148:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  148 |     virtual value & at(const std::string & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:149:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  149 |     virtual value & at(int64_t /* idx */, value & /* default_val */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:150:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  150 |     virtual value & at(int64_t /* idx */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.cpp:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:110:19: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  110 |     virtual value execute_impl(context &) { throw std::runtime_error("cannot exec " + type()); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:146:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:198:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  198 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:212:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  212 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:504:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  504 |     value execute_impl(context &) override {
      |           ^
21 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.cpp:2:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:130:21: warning: function 'as_int' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  130 |     virtual int64_t as_int() const { throw std::runtime_error(type() + " is not an int value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:131:20: warning: function 'as_float' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  131 |     virtual double as_float() const { throw std::runtime_error(type() + " is not a float value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:132:20: warning: function 'as_string' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  132 |     virtual string as_string() const { throw std::runtime_error(type() + " is not a string value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:133:18: warning: function 'as_bool' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  133 |     virtual bool as_bool() const { throw std::runtime_error(type() + " is not a bool value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:134:40: warning: function 'as_array' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  134 |     virtual const std::vector<value> & as_array() const { throw std::runtime_error(type() + " is not an array value"); }
      |                                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:135:58: warning: function 'as_ordered_object' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  135 |     virtual const std::vector<std::pair<value, value>> & as_ordered_object() const { throw std::runtime_error(type() + " is not an object value"); }
      |                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:136:19: warning: function 'invoke' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  136 |     virtual value invoke(const func_args &) const { throw std::runtime_error(type() + " is not a function value"); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:139:35: warning: function 'get_builtins' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  139 |     virtual const func_builtins & get_builtins() const {
      |                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:143:18: warning: function 'has_key' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  143 |     virtual bool has_key(const value &) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:144:18: warning: function 'insert' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  144 |     virtual void insert(const value & /* key */, const value & /* val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:145:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  145 |     virtual value & at(const value & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:146:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     virtual value & at(const value & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:147:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  147 |     virtual value & at(const std::string & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:148:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  148 |     virtual value & at(const std::string & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:149:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  149 |     virtual value & at(int64_t /* idx */, value & /* default_val */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:150:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  150 |     virtual value & at(int64_t /* idx */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.cpp:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:110:19: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  110 |     virtual value execute_impl(context &) { throw std::runtime_error("cannot exec " + type()); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:146:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:198:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  198 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:212:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  212 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:504:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  504 |     value execute_impl(context &) override {
      |           ^
21 warnings generated when compiling for host.
[346/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/jinja/string.cpp.o -MF common/CMakeFiles/common.dir/jinja/string.cpp.o.d -o common/CMakeFiles/common.dir/jinja/string.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/string.cpp
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/string.cpp:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:130:21: warning: function 'as_int' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  130 |     virtual int64_t as_int() const { throw std::runtime_error(type() + " is not an int value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:131:20: warning: function 'as_float' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  131 |     virtual double as_float() const { throw std::runtime_error(type() + " is not a float value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:132:20: warning: function 'as_string' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  132 |     virtual string as_string() const { throw std::runtime_error(type() + " is not a string value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:133:18: warning: function 'as_bool' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  133 |     virtual bool as_bool() const { throw std::runtime_error(type() + " is not a bool value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:134:40: warning: function 'as_array' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  134 |     virtual const std::vector<value> & as_array() const { throw std::runtime_error(type() + " is not an array value"); }
      |                                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:135:58: warning: function 'as_ordered_object' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  135 |     virtual const std::vector<std::pair<value, value>> & as_ordered_object() const { throw std::runtime_error(type() + " is not an object value"); }
      |                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:136:19: warning: function 'invoke' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  136 |     virtual value invoke(const func_args &) const { throw std::runtime_error(type() + " is not a function value"); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:139:35: warning: function 'get_builtins' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  139 |     virtual const func_builtins & get_builtins() const {
      |                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:143:18: warning: function 'has_key' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  143 |     virtual bool has_key(const value &) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:144:18: warning: function 'insert' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  144 |     virtual void insert(const value & /* key */, const value & /* val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:145:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  145 |     virtual value & at(const value & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:146:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     virtual value & at(const value & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:147:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  147 |     virtual value & at(const std::string & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:148:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  148 |     virtual value & at(const std::string & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:149:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  149 |     virtual value & at(int64_t /* idx */, value & /* default_val */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:150:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  150 |     virtual value & at(int64_t /* idx */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
16 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/string.cpp:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:130:21: warning: function 'as_int' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  130 |     virtual int64_t as_int() const { throw std::runtime_error(type() + " is not an int value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:131:20: warning: function 'as_float' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  131 |     virtual double as_float() const { throw std::runtime_error(type() + " is not a float value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:132:20: warning: function 'as_string' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  132 |     virtual string as_string() const { throw std::runtime_error(type() + " is not a string value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:133:18: warning: function 'as_bool' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  133 |     virtual bool as_bool() const { throw std::runtime_error(type() + " is not a bool value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:134:40: warning: function 'as_array' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  134 |     virtual const std::vector<value> & as_array() const { throw std::runtime_error(type() + " is not an array value"); }
      |                                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:135:58: warning: function 'as_ordered_object' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  135 |     virtual const std::vector<std::pair<value, value>> & as_ordered_object() const { throw std::runtime_error(type() + " is not an object value"); }
      |                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:136:19: warning: function 'invoke' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  136 |     virtual value invoke(const func_args &) const { throw std::runtime_error(type() + " is not a function value"); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:139:35: warning: function 'get_builtins' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  139 |     virtual const func_builtins & get_builtins() const {
      |                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:143:18: warning: function 'has_key' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  143 |     virtual bool has_key(const value &) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:144:18: warning: function 'insert' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  144 |     virtual void insert(const value & /* key */, const value & /* val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:145:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  145 |     virtual value & at(const value & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:146:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     virtual value & at(const value & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:147:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  147 |     virtual value & at(const std::string & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:148:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  148 |     virtual value & at(const std::string & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:149:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  149 |     virtual value & at(int64_t /* idx */, value & /* default_val */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/./jinja/value.h:150:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  150 |     virtual value & at(int64_t /* idx */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
16 warnings generated when compiling for host.
[347/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/__/license.cpp.o -MF common/CMakeFiles/common.dir/__/license.cpp.o.d -o common/CMakeFiles/common.dir/__/license.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/redhat-linux-build/license.cpp
[348/401] : && /usr/bin/cmake -E rm -f vendor/cpp-httplib/libcpp-httplib.a && /usr/bin/ar qc vendor/cpp-httplib/libcpp-httplib.a  vendor/cpp-httplib/CMakeFiles/cpp-httplib.dir/httplib.cpp.o && /usr/bin/ranlib vendor/cpp-httplib/libcpp-httplib.a && :
[349/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/peg-parser.cpp.o -MF common/CMakeFiles/common.dir/peg-parser.cpp.o.d -o common/CMakeFiles/common.dir/peg-parser.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/peg-parser.cpp
[350/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/jinja/value.cpp.o -MF common/CMakeFiles/common.dir/jinja/value.cpp.o.d -o common/CMakeFiles/common.dir/jinja/value.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.cpp
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.cpp:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:130:21: warning: function 'as_int' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  130 |     virtual int64_t as_int() const { throw std::runtime_error(type() + " is not an int value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:131:20: warning: function 'as_float' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  131 |     virtual double as_float() const { throw std::runtime_error(type() + " is not a float value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:132:20: warning: function 'as_string' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  132 |     virtual string as_string() const { throw std::runtime_error(type() + " is not a string value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:133:18: warning: function 'as_bool' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  133 |     virtual bool as_bool() const { throw std::runtime_error(type() + " is not a bool value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:134:40: warning: function 'as_array' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  134 |     virtual const std::vector<value> & as_array() const { throw std::runtime_error(type() + " is not an array value"); }
      |                                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:135:58: warning: function 'as_ordered_object' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  135 |     virtual const std::vector<std::pair<value, value>> & as_ordered_object() const { throw std::runtime_error(type() + " is not an object value"); }
      |                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:136:19: warning: function 'invoke' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  136 |     virtual value invoke(const func_args &) const { throw std::runtime_error(type() + " is not a function value"); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:139:35: warning: function 'get_builtins' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  139 |     virtual const func_builtins & get_builtins() const {
      |                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:143:18: warning: function 'has_key' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  143 |     virtual bool has_key(const value &) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:144:18: warning: function 'insert' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  144 |     virtual void insert(const value & /* key */, const value & /* val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:145:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  145 |     virtual value & at(const value & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:146:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     virtual value & at(const value & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:147:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  147 |     virtual value & at(const std::string & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:148:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  148 |     virtual value & at(const std::string & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:149:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  149 |     virtual value & at(int64_t /* idx */, value & /* default_val */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:150:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  150 |     virtual value & at(int64_t /* idx */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.cpp:1:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:110:19: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  110 |     virtual value execute_impl(context &) { throw std::runtime_error("cannot exec " + type()); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:146:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:198:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  198 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:212:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  212 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:504:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  504 |     value execute_impl(context &) override {
      |           ^
21 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.cpp:1:
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:4:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:130:21: warning: function 'as_int' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  130 |     virtual int64_t as_int() const { throw std::runtime_error(type() + " is not an int value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:131:20: warning: function 'as_float' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  131 |     virtual double as_float() const { throw std::runtime_error(type() + " is not a float value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:132:20: warning: function 'as_string' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  132 |     virtual string as_string() const { throw std::runtime_error(type() + " is not a string value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:133:18: warning: function 'as_bool' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  133 |     virtual bool as_bool() const { throw std::runtime_error(type() + " is not a bool value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:134:40: warning: function 'as_array' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  134 |     virtual const std::vector<value> & as_array() const { throw std::runtime_error(type() + " is not an array value"); }
      |                                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:135:58: warning: function 'as_ordered_object' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  135 |     virtual const std::vector<std::pair<value, value>> & as_ordered_object() const { throw std::runtime_error(type() + " is not an object value"); }
      |                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:136:19: warning: function 'invoke' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  136 |     virtual value invoke(const func_args &) const { throw std::runtime_error(type() + " is not a function value"); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:139:35: warning: function 'get_builtins' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  139 |     virtual const func_builtins & get_builtins() const {
      |                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:143:18: warning: function 'has_key' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  143 |     virtual bool has_key(const value &) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:144:18: warning: function 'insert' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  144 |     virtual void insert(const value & /* key */, const value & /* val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:145:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  145 |     virtual value & at(const value & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:146:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     virtual value & at(const value & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:147:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  147 |     virtual value & at(const std::string & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:148:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  148 |     virtual value & at(const std::string & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:149:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  149 |     virtual value & at(int64_t /* idx */, value & /* default_val */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:150:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  150 |     virtual value & at(int64_t /* idx */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.cpp:1:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:110:19: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  110 |     virtual value execute_impl(context &) { throw std::runtime_error("cannot exec " + type()); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:146:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:198:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  198 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:212:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  212 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:504:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  504 |     value execute_impl(context &) override {
      |           ^
21 warnings generated when compiling for host.
[351/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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/jinja/caps.cpp.o -MF common/CMakeFiles/common.dir/jinja/caps.cpp.o.d -o common/CMakeFiles/common.dir/jinja/caps.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/caps.cpp
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/caps.cpp:1:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:130:21: warning: function 'as_int' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  130 |     virtual int64_t as_int() const { throw std::runtime_error(type() + " is not an int value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:131:20: warning: function 'as_float' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  131 |     virtual double as_float() const { throw std::runtime_error(type() + " is not a float value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:132:20: warning: function 'as_string' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  132 |     virtual string as_string() const { throw std::runtime_error(type() + " is not a string value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:133:18: warning: function 'as_bool' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  133 |     virtual bool as_bool() const { throw std::runtime_error(type() + " is not a bool value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:134:40: warning: function 'as_array' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  134 |     virtual const std::vector<value> & as_array() const { throw std::runtime_error(type() + " is not an array value"); }
      |                                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:135:58: warning: function 'as_ordered_object' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  135 |     virtual const std::vector<std::pair<value, value>> & as_ordered_object() const { throw std::runtime_error(type() + " is not an object value"); }
      |                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:136:19: warning: function 'invoke' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  136 |     virtual value invoke(const func_args &) const { throw std::runtime_error(type() + " is not a function value"); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:139:35: warning: function 'get_builtins' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  139 |     virtual const func_builtins & get_builtins() const {
      |                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:143:18: warning: function 'has_key' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  143 |     virtual bool has_key(const value &) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:144:18: warning: function 'insert' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  144 |     virtual void insert(const value & /* key */, const value & /* val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:145:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  145 |     virtual value & at(const value & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:146:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     virtual value & at(const value & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:147:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  147 |     virtual value & at(const std::string & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:148:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  148 |     virtual value & at(const std::string & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:149:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  149 |     virtual value & at(int64_t /* idx */, value & /* default_val */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:150:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  150 |     virtual value & at(int64_t /* idx */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/caps.cpp:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:110:19: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  110 |     virtual value execute_impl(context &) { throw std::runtime_error("cannot exec " + type()); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:146:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:198:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  198 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:212:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  212 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:504:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  504 |     value execute_impl(context &) override {
      |           ^
21 warnings generated when compiling for gfx906.
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/caps.cpp:1:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:130:21: warning: function 'as_int' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  130 |     virtual int64_t as_int() const { throw std::runtime_error(type() + " is not an int value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:131:20: warning: function 'as_float' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  131 |     virtual double as_float() const { throw std::runtime_error(type() + " is not a float value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:132:20: warning: function 'as_string' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  132 |     virtual string as_string() const { throw std::runtime_error(type() + " is not a string value"); }
      |                    ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:133:18: warning: function 'as_bool' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  133 |     virtual bool as_bool() const { throw std::runtime_error(type() + " is not a bool value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:134:40: warning: function 'as_array' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  134 |     virtual const std::vector<value> & as_array() const { throw std::runtime_error(type() + " is not an array value"); }
      |                                        ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:135:58: warning: function 'as_ordered_object' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  135 |     virtual const std::vector<std::pair<value, value>> & as_ordered_object() const { throw std::runtime_error(type() + " is not an object value"); }
      |                                                          ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:136:19: warning: function 'invoke' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  136 |     virtual value invoke(const func_args &) const { throw std::runtime_error(type() + " is not a function value"); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:139:35: warning: function 'get_builtins' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  139 |     virtual const func_builtins & get_builtins() const {
      |                                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:143:18: warning: function 'has_key' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  143 |     virtual bool has_key(const value &) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:144:18: warning: function 'insert' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  144 |     virtual void insert(const value & /* key */, const value & /* val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                  ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:145:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  145 |     virtual value & at(const value & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:146:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     virtual value & at(const value & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:147:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  147 |     virtual value & at(const std::string & /* key */, value & /* default_val */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:148:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  148 |     virtual value & at(const std::string & /* key */) { throw std::runtime_error(type() + " is not an object value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:149:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  149 |     virtual value & at(int64_t /* idx */, value & /* default_val */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/value.h:150:21: warning: function 'at' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  150 |     virtual value & at(int64_t /* idx */) { throw std::runtime_error(type() + " is not an array value"); }
      |                     ^
In file included from /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/caps.cpp:2:
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:110:19: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  110 |     virtual value execute_impl(context &) { throw std::runtime_error("cannot exec " + type()); }
      |                   ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:146:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  146 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:198:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  198 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:212:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  212 |     value execute_impl(context &) override {
      |           ^
/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/jinja/runtime.h:504:11: warning: function 'execute_impl' could be declared with attribute 'noreturn' [-Wmissing-noreturn]
  504 |     value execute_impl(context &) override {
      |           ^
21 warnings generated when compiling for host.
[352/401] : && /usr/bin/cmake -E rm -f common/libcommon.a && /usr/bin/ar qc common/libcommon.a  common/CMakeFiles/build_info.dir/build-info.cpp.o common/CMakeFiles/common.dir/arg.cpp.o common/CMakeFiles/common.dir/chat-parser.cpp.o common/CMakeFiles/common.dir/chat-parser-xml-toolcall.cpp.o common/CMakeFiles/common.dir/chat-peg-parser.cpp.o common/CMakeFiles/common.dir/chat.cpp.o common/CMakeFiles/common.dir/common.cpp.o common/CMakeFiles/common.dir/console.cpp.o common/CMakeFiles/common.dir/debug.cpp.o common/CMakeFiles/common.dir/download.cpp.o common/CMakeFiles/common.dir/json-partial.cpp.o common/CMakeFiles/common.dir/json-schema-to-grammar.cpp.o common/CMakeFiles/common.dir/llguidance.cpp.o common/CMakeFiles/common.dir/log.cpp.o common/CMakeFiles/common.dir/ngram-cache.cpp.o common/CMakeFiles/common.dir/ngram-map.cpp.o common/CMakeFiles/common.dir/ngram-mod.cpp.o common/CMakeFiles/common.dir/peg-parser.cpp.o common/CMakeFiles/common.dir/preset.cpp.o common/CMakeFiles/common.dir/regex-partial.cpp.o common/CMakeFiles/common.dir/sampling.cpp.o common/CMakeFiles/common.dir/speculative.cpp.o common/CMakeFiles/common.dir/unicode.cpp.o common/CMakeFiles/common.dir/jinja/lexer.cpp.o common/CMakeFiles/common.dir/jinja/parser.cpp.o common/CMakeFiles/common.dir/jinja/runtime.cpp.o common/CMakeFiles/common.dir/jinja/value.cpp.o common/CMakeFiles/common.dir/jinja/string.cpp.o common/CMakeFiles/common.dir/jinja/caps.cpp.o common/CMakeFiles/common.dir/__/license.cpp.o && /usr/bin/ranlib common/libcommon.a && :
[353/401] : && /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  -shared -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 -Xlinker --dependency-file=tools/mtmd/CMakeFiles/mtmd.dir/link.d -Wl,-soname,libmtmd.so.0 -o bin/libmtmd.so.0.0.0 tools/mtmd/CMakeFiles/mtmd.dir/mtmd.cpp.o tools/mtmd/CMakeFiles/mtmd.dir/mtmd-audio.cpp.o tools/mtmd/CMakeFiles/mtmd.dir/mtmd-helper.cpp.o tools/mtmd/CMakeFiles/mtmd.dir/clip.cpp.o tools/mtmd/CMakeFiles/mtmd.dir/models/cogvlm.cpp.o tools/mtmd/CMakeFiles/mtmd.dir/models/conformer.cpp.o tools/mtmd/CMakeFiles/mtmd.dir/models/glm4v.cpp.o tools/mtmd/CMakeFiles/mtmd.dir/models/internvl.cpp.o tools/mtmd/CMakeFiles/mtmd.dir/models/kimivl.cpp.o tools/mtmd/CMakeFiles/mtmd.dir/models/kimik25.cpp.o tools/mtmd/CMakeFiles/mtmd.dir/models/nemotron-v2-vl.cpp.o tools/mtmd/CMakeFiles/mtmd.dir/models/llama4.cpp.o tools/mtmd/CMakeFiles/mtmd.dir/models/llava.cpp.o tools/mtmd/CMakeFiles/mtmd.dir/models/minicpmv.cpp.o tools/mtmd/CMakeFiles/mtmd.dir/models/pixtral.cpp.o tools/mtmd/CMakeFiles/mtmd.dir/models/qwen2vl.cpp.o tools/mtmd/CMakeFiles/mtmd.dir/models/qwen3vl.cpp.o tools/mtmd/CMakeFiles/mtmd.dir/models/siglip.cpp.o tools/mtmd/CMakeFiles/mtmd.dir/models/whisper-enc.cpp.o tools/mtmd/CMakeFiles/mtmd.dir/models/mobilenetv5.cpp.o tools/mtmd/CMakeFiles/mtmd.dir/models/youtuvl.cpp.o  bin/libllama.so.0.0.0  bin/libggml.so.0.9.5  bin/libggml-cpu.so.0.9.5  bin/libggml-hip.so.0.9.5  bin/libggml-base.so.0.9.5 && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
[354/401] /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-b8064-build/llama.cpp-b8064/tools/server -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/redhat-linux-build/tools/server -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/server/../mtmd -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064 -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -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 tools/server/CMakeFiles/server-context.dir/server-queue.cpp.o -MF tools/server/CMakeFiles/server-context.dir/server-queue.cpp.o.d -o tools/server/CMakeFiles/server-context.dir/server-queue.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/server/server-queue.cpp
[355/401] /usr/bin/cmake -E cmake_symlink_library bin/libmtmd.so.0.0.0  bin/libmtmd.so.0 bin/libmtmd.so && :
[356/401] /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-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT tools/batched-bench/CMakeFiles/llama-batched-bench.dir/batched-bench.cpp.o -MF tools/batched-bench/CMakeFiles/llama-batched-bench.dir/batched-bench.cpp.o.d -o tools/batched-bench/CMakeFiles/llama-batched-bench.dir/batched-bench.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/batched-bench/batched-bench.cpp
[357/401] /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-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT tools/gguf-split/CMakeFiles/llama-gguf-split.dir/gguf-split.cpp.o -MF tools/gguf-split/CMakeFiles/llama-gguf-split.dir/gguf-split.cpp.o.d -o tools/gguf-split/CMakeFiles/llama-gguf-split.dir/gguf-split.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/gguf-split/gguf-split.cpp
[358/401] /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-b8064-build/llama.cpp-b8064/tools/server -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/redhat-linux-build/tools/server -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/server/../mtmd -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064 -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -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 tools/server/CMakeFiles/server-context.dir/server-common.cpp.o -MF tools/server/CMakeFiles/server-context.dir/server-common.cpp.o.d -o tools/server/CMakeFiles/server-context.dir/server-common.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/server/server-common.cpp
[359/401] /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-b8064-build/llama.cpp-b8064/tools/server -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/redhat-linux-build/tools/server -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/server/../mtmd -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064 -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -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 tools/server/CMakeFiles/server-context.dir/server-task.cpp.o -MF tools/server/CMakeFiles/server-context.dir/server-task.cpp.o.d -o tools/server/CMakeFiles/server-context.dir/server-task.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/server/server-task.cpp
[360/401] /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-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT tools/completion/CMakeFiles/llama-completion.dir/completion.cpp.o -MF tools/completion/CMakeFiles/llama-completion.dir/completion.cpp.o.d -o tools/completion/CMakeFiles/llama-completion.dir/completion.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/completion/completion.cpp
[361/401] /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-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT tools/imatrix/CMakeFiles/llama-imatrix.dir/imatrix.cpp.o -MF tools/imatrix/CMakeFiles/llama-imatrix.dir/imatrix.cpp.o.d -o tools/imatrix/CMakeFiles/llama-imatrix.dir/imatrix.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/imatrix/imatrix.cpp
[362/401] /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-b8064-build/llama.cpp-b8064/tools/server -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/redhat-linux-build/tools/server -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/server/../mtmd -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064 -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -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 tools/server/CMakeFiles/server-context.dir/server-context.cpp.o -MF tools/server/CMakeFiles/server-context.dir/server-context.cpp.o.d -o tools/server/CMakeFiles/server-context.dir/server-context.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/server/server-context.cpp
[363/401] : && /usr/bin/cmake -E rm -f tools/server/libserver-context.a && /usr/bin/ar qc tools/server/libserver-context.a  tools/server/CMakeFiles/server-context.dir/server-task.cpp.o tools/server/CMakeFiles/server-context.dir/server-queue.cpp.o tools/server/CMakeFiles/server-context.dir/server-common.cpp.o tools/server/CMakeFiles/server-context.dir/server-context.cpp.o && /usr/bin/ranlib tools/server/libserver-context.a && :
[364/401] /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-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT tools/llama-bench/CMakeFiles/llama-bench.dir/llama-bench.cpp.o -MF tools/llama-bench/CMakeFiles/llama-bench.dir/llama-bench.cpp.o.d -o tools/llama-bench/CMakeFiles/llama-bench.dir/llama-bench.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/llama-bench/llama-bench.cpp
[365/401] cd /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/redhat-linux-build/tools/server && /usr/bin/cmake -DINPUT=/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/server/public/loading.html -DOUTPUT=/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/redhat-linux-build/tools/server/loading.html.hpp -P /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/scripts/xxd.cmake
[366/401] cd /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/redhat-linux-build/tools/server && /usr/bin/cmake -DINPUT=/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/server/public/index.html.gz -DOUTPUT=/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/redhat-linux-build/tools/server/index.html.gz.hpp -P /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/scripts/xxd.cmake
[367/401] /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-b8064-build/llama.cpp-b8064/tools/quantize/../../common -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT tools/quantize/CMakeFiles/llama-quantize.dir/quantize.cpp.o -MF tools/quantize/CMakeFiles/llama-quantize.dir/quantize.cpp.o.d -o tools/quantize/CMakeFiles/llama-quantize.dir/quantize.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/quantize/quantize.cpp
[368/401] /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-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT tools/perplexity/CMakeFiles/llama-perplexity.dir/perplexity.cpp.o -MF tools/perplexity/CMakeFiles/llama-perplexity.dir/perplexity.cpp.o.d -o tools/perplexity/CMakeFiles/llama-perplexity.dir/perplexity.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/perplexity/perplexity.cpp
[369/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/server -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/redhat-linux-build/tools/server -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/server/../mtmd -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064 -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -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 -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT tools/server/CMakeFiles/llama-server.dir/server.cpp.o -MF tools/server/CMakeFiles/llama-server.dir/server.cpp.o.d -o tools/server/CMakeFiles/llama-server.dir/server.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/server/server.cpp
[370/401] /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-b8064-build/llama.cpp-b8064/tools/cli/../server -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -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 -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT tools/cli/CMakeFiles/llama-cli.dir/cli.cpp.o -MF tools/cli/CMakeFiles/llama-cli.dir/cli.cpp.o.d -o tools/cli/CMakeFiles/llama-cli.dir/cli.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/cli/cli.cpp
[371/401] /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-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT tools/tokenize/CMakeFiles/llama-tokenize.dir/tokenize.cpp.o -MF tools/tokenize/CMakeFiles/llama-tokenize.dir/tokenize.cpp.o.d -o tools/tokenize/CMakeFiles/llama-tokenize.dir/tokenize.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/tokenize/tokenize.cpp
[372/401] /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 -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT tools/mtmd/CMakeFiles/llama-llava-cli.dir/deprecation-warning.cpp.o -MF tools/mtmd/CMakeFiles/llama-llava-cli.dir/deprecation-warning.cpp.o.d -o tools/mtmd/CMakeFiles/llama-llava-cli.dir/deprecation-warning.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/deprecation-warning.cpp
[373/401] /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 -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT tools/mtmd/CMakeFiles/llama-gemma3-cli.dir/deprecation-warning.cpp.o -MF tools/mtmd/CMakeFiles/llama-gemma3-cli.dir/deprecation-warning.cpp.o.d -o tools/mtmd/CMakeFiles/llama-gemma3-cli.dir/deprecation-warning.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/deprecation-warning.cpp
[374/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/server -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/redhat-linux-build/tools/server -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/server/../mtmd -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064 -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -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 -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT tools/server/CMakeFiles/llama-server.dir/server-http.cpp.o -MF tools/server/CMakeFiles/llama-server.dir/server-http.cpp.o.d -o tools/server/CMakeFiles/llama-server.dir/server-http.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/server/server-http.cpp
[375/401] /usr/bin/hipcc -DCPPHTTPLIB_OPENSSL_SUPPORT -DGGML_BACKEND_SHARED -DGGML_SHARED -DGGML_USE_CPU -DGGML_USE_CUDA -DGGML_USE_HIP -DLLAMA_SHARED -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/server -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/redhat-linux-build/tools/server -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/server/../mtmd -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064 -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -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 -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT tools/server/CMakeFiles/llama-server.dir/server-models.cpp.o -MF tools/server/CMakeFiles/llama-server.dir/server-models.cpp.o.d -o tools/server/CMakeFiles/llama-server.dir/server-models.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/server/server-models.cpp
[376/401] /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 -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT tools/mtmd/CMakeFiles/llama-minicpmv-cli.dir/deprecation-warning.cpp.o -MF tools/mtmd/CMakeFiles/llama-minicpmv-cli.dir/deprecation-warning.cpp.o.d -o tools/mtmd/CMakeFiles/llama-minicpmv-cli.dir/deprecation-warning.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/deprecation-warning.cpp
[377/401] /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 -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT tools/mtmd/CMakeFiles/llama-qwen2vl-cli.dir/deprecation-warning.cpp.o -MF tools/mtmd/CMakeFiles/llama-qwen2vl-cli.dir/deprecation-warning.cpp.o.d -o tools/mtmd/CMakeFiles/llama-qwen2vl-cli.dir/deprecation-warning.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/deprecation-warning.cpp
[378/401] /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-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT tools/export-lora/CMakeFiles/llama-export-lora.dir/export-lora.cpp.o -MF tools/export-lora/CMakeFiles/llama-export-lora.dir/export-lora.cpp.o.d -o tools/export-lora/CMakeFiles/llama-export-lora.dir/export-lora.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/export-lora/export-lora.cpp
[379/401] /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-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT tools/cvector-generator/CMakeFiles/llama-cvector-generator.dir/cvector-generator.cpp.o -MF tools/cvector-generator/CMakeFiles/llama-cvector-generator.dir/cvector-generator.cpp.o.d -o tools/cvector-generator/CMakeFiles/llama-cvector-generator.dir/cvector-generator.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/cvector-generator/cvector-generator.cpp
[380/401] /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-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/. -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 -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT tools/mtmd/CMakeFiles/llama-mtmd-cli.dir/mtmd-cli.cpp.o -MF tools/mtmd/CMakeFiles/llama-mtmd-cli.dir/mtmd-cli.cpp.o.d -o tools/mtmd/CMakeFiles/llama-mtmd-cli.dir/mtmd-cli.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/mtmd/mtmd-cli.cpp
[381/401] : && /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 -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 -Xlinker --dependency-file=tools/gguf-split/CMakeFiles/llama-gguf-split.dir/link.d tools/gguf-split/CMakeFiles/llama-gguf-split.dir/gguf-split.cpp.o -o bin/llama-gguf-split  common/libcommon.a  bin/libllama.so.0.0.0  bin/libggml.so.0.9.5  bin/libggml-cpu.so.0.9.5  bin/libggml-hip.so.0.9.5  bin/libggml-base.so.0.9.5  vendor/cpp-httplib/libcpp-httplib.a  /usr/lib64/libssl.so  /usr/lib64/libcrypto.so && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
[382/401] /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-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/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 -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT tools/fit-params/CMakeFiles/llama-fit-params.dir/fit-params.cpp.o -MF tools/fit-params/CMakeFiles/llama-fit-params.dir/fit-params.cpp.o.d -o tools/fit-params/CMakeFiles/llama-fit-params.dir/fit-params.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/fit-params/fit-params.cpp
[383/401] /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-b8064-build/llama.cpp-b8064/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/ggml/src/../include -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/. -I/builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/common/../vendor -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 -Wmissing-declarations -Wmissing-noreturn -Wall -Wextra -Wpedantic -Wcast-qual -Wno-unused-function -Wunreachable-code-break -Wunreachable-code-return -Wmissing-prototypes -Wextra-semi -MD -MT tools/tts/CMakeFiles/llama-tts.dir/tts.cpp.o -MF tools/tts/CMakeFiles/llama-tts.dir/tts.cpp.o.d -o tools/tts/CMakeFiles/llama-tts.dir/tts.cpp.o -c /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/tools/tts/tts.cpp
[384/401] : && /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 -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 -Xlinker --dependency-file=tools/llama-bench/CMakeFiles/llama-bench.dir/link.d tools/llama-bench/CMakeFiles/llama-bench.dir/llama-bench.cpp.o -o bin/llama-bench  common/libcommon.a  bin/libllama.so.0.0.0  bin/libggml.so.0.9.5  bin/libggml-cpu.so.0.9.5  bin/libggml-hip.so.0.9.5  bin/libggml-base.so.0.9.5  vendor/cpp-httplib/libcpp-httplib.a  /usr/lib64/libssl.so  /usr/lib64/libcrypto.so && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
[385/401] : && /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 -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 -Xlinker --dependency-file=tools/batched-bench/CMakeFiles/llama-batched-bench.dir/link.d tools/batched-bench/CMakeFiles/llama-batched-bench.dir/batched-bench.cpp.o -o bin/llama-batched-bench  common/libcommon.a  bin/libllama.so.0.0.0  bin/libggml.so.0.9.5  bin/libggml-cpu.so.0.9.5  bin/libggml-hip.so.0.9.5  bin/libggml-base.so.0.9.5  vendor/cpp-httplib/libcpp-httplib.a  /usr/lib64/libssl.so  /usr/lib64/libcrypto.so && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
[386/401] : && /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 -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 -Xlinker --dependency-file=tools/quantize/CMakeFiles/llama-quantize.dir/link.d tools/quantize/CMakeFiles/llama-quantize.dir/quantize.cpp.o -o bin/llama-quantize  common/libcommon.a  bin/libllama.so.0.0.0  bin/libggml.so.0.9.5  bin/libggml-cpu.so.0.9.5  bin/libggml-hip.so.0.9.5  bin/libggml-base.so.0.9.5  vendor/cpp-httplib/libcpp-httplib.a  /usr/lib64/libssl.so  /usr/lib64/libcrypto.so && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
[387/401] : && /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 -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 -Xlinker --dependency-file=tools/imatrix/CMakeFiles/llama-imatrix.dir/link.d tools/imatrix/CMakeFiles/llama-imatrix.dir/imatrix.cpp.o -o bin/llama-imatrix  common/libcommon.a  bin/libllama.so.0.0.0  bin/libggml.so.0.9.5  bin/libggml-cpu.so.0.9.5  bin/libggml-hip.so.0.9.5  bin/libggml-base.so.0.9.5  vendor/cpp-httplib/libcpp-httplib.a  /usr/lib64/libssl.so  /usr/lib64/libcrypto.so && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
[388/401] : && /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 -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 -Xlinker --dependency-file=tools/completion/CMakeFiles/llama-completion.dir/link.d tools/completion/CMakeFiles/llama-completion.dir/completion.cpp.o -o bin/llama-completion  common/libcommon.a  bin/libllama.so.0.0.0  bin/libggml.so.0.9.5  bin/libggml-cpu.so.0.9.5  bin/libggml-hip.so.0.9.5  bin/libggml-base.so.0.9.5  vendor/cpp-httplib/libcpp-httplib.a  /usr/lib64/libssl.so  /usr/lib64/libcrypto.so && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
[389/401] : && /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 -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 -Xlinker --dependency-file=tools/tokenize/CMakeFiles/llama-tokenize.dir/link.d tools/tokenize/CMakeFiles/llama-tokenize.dir/tokenize.cpp.o -o bin/llama-tokenize  common/libcommon.a  bin/libllama.so.0.0.0  bin/libggml.so.0.9.5  bin/libggml-cpu.so.0.9.5  bin/libggml-hip.so.0.9.5  bin/libggml-base.so.0.9.5  vendor/cpp-httplib/libcpp-httplib.a  /usr/lib64/libssl.so  /usr/lib64/libcrypto.so && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
[390/401] : && /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 -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 -Xlinker --dependency-file=tools/perplexity/CMakeFiles/llama-perplexity.dir/link.d tools/perplexity/CMakeFiles/llama-perplexity.dir/perplexity.cpp.o -o bin/llama-perplexity  common/libcommon.a  bin/libllama.so.0.0.0  bin/libggml.so.0.9.5  bin/libggml-cpu.so.0.9.5  bin/libggml-hip.so.0.9.5  bin/libggml-base.so.0.9.5  vendor/cpp-httplib/libcpp-httplib.a  /usr/lib64/libssl.so  /usr/lib64/libcrypto.so && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
[391/401] : && /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 -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 -Xlinker --dependency-file=tools/mtmd/CMakeFiles/llama-llava-cli.dir/link.d tools/mtmd/CMakeFiles/llama-llava-cli.dir/deprecation-warning.cpp.o -o bin/llama-llava-cli   && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
[392/401] : && /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 -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 -Xlinker --dependency-file=tools/mtmd/CMakeFiles/llama-gemma3-cli.dir/link.d tools/mtmd/CMakeFiles/llama-gemma3-cli.dir/deprecation-warning.cpp.o -o bin/llama-gemma3-cli   && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
[393/401] : && /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 -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 -Xlinker --dependency-file=tools/mtmd/CMakeFiles/llama-minicpmv-cli.dir/link.d tools/mtmd/CMakeFiles/llama-minicpmv-cli.dir/deprecation-warning.cpp.o -o bin/llama-minicpmv-cli   && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
[394/401] : && /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 -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 -Xlinker --dependency-file=tools/mtmd/CMakeFiles/llama-qwen2vl-cli.dir/link.d tools/mtmd/CMakeFiles/llama-qwen2vl-cli.dir/deprecation-warning.cpp.o -o bin/llama-qwen2vl-cli   && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
[395/401] : && /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 -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 -Xlinker --dependency-file=tools/tts/CMakeFiles/llama-tts.dir/link.d tools/tts/CMakeFiles/llama-tts.dir/tts.cpp.o -o bin/llama-tts  common/libcommon.a  bin/libllama.so.0.0.0  bin/libggml.so.0.9.5  bin/libggml-cpu.so.0.9.5  bin/libggml-hip.so.0.9.5  bin/libggml-base.so.0.9.5  vendor/cpp-httplib/libcpp-httplib.a  /usr/lib64/libssl.so  /usr/lib64/libcrypto.so && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
[396/401] : && /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 -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 -Xlinker --dependency-file=tools/mtmd/CMakeFiles/llama-mtmd-cli.dir/link.d tools/mtmd/CMakeFiles/llama-mtmd-cli.dir/mtmd-cli.cpp.o -o bin/llama-mtmd-cli  common/libcommon.a  bin/libmtmd.so.0.0.0  vendor/cpp-httplib/libcpp-httplib.a  /usr/lib64/libssl.so  /usr/lib64/libcrypto.so  bin/libllama.so.0.0.0  bin/libggml.so.0.9.5  bin/libggml-cpu.so.0.9.5  bin/libggml-hip.so.0.9.5  bin/libggml-base.so.0.9.5 && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
[397/401] : && /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 -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 -Xlinker --dependency-file=tools/cli/CMakeFiles/llama-cli.dir/link.d tools/cli/CMakeFiles/llama-cli.dir/cli.cpp.o -o bin/llama-cli  tools/server/libserver-context.a  common/libcommon.a  vendor/cpp-httplib/libcpp-httplib.a  /usr/lib64/libssl.so  /usr/lib64/libcrypto.so  bin/libmtmd.so.0.0.0  bin/libllama.so.0.0.0  bin/libggml.so.0.9.5  bin/libggml-cpu.so.0.9.5  bin/libggml-hip.so.0.9.5  bin/libggml-base.so.0.9.5 && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
[398/401] : && /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 -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 -Xlinker --dependency-file=tools/server/CMakeFiles/llama-server.dir/link.d tools/server/CMakeFiles/llama-server.dir/server.cpp.o tools/server/CMakeFiles/llama-server.dir/server-http.cpp.o tools/server/CMakeFiles/llama-server.dir/server-models.cpp.o -o bin/llama-server  tools/server/libserver-context.a  common/libcommon.a  vendor/cpp-httplib/libcpp-httplib.a  /usr/lib64/libssl.so  /usr/lib64/libcrypto.so  bin/libmtmd.so.0.0.0  bin/libllama.so.0.0.0  bin/libggml.so.0.9.5  bin/libggml-cpu.so.0.9.5  bin/libggml-hip.so.0.9.5  bin/libggml-base.so.0.9.5 && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
[399/401] : && /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 -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 -Xlinker --dependency-file=tools/cvector-generator/CMakeFiles/llama-cvector-generator.dir/link.d tools/cvector-generator/CMakeFiles/llama-cvector-generator.dir/cvector-generator.cpp.o -o bin/llama-cvector-generator  common/libcommon.a  bin/libllama.so.0.0.0  bin/libggml.so.0.9.5  bin/libggml-cpu.so.0.9.5  bin/libggml-hip.so.0.9.5  bin/libggml-base.so.0.9.5  vendor/cpp-httplib/libcpp-httplib.a  /usr/lib64/libssl.so  /usr/lib64/libcrypto.so && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
[400/401] : && /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 -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 -Xlinker --dependency-file=tools/export-lora/CMakeFiles/llama-export-lora.dir/link.d tools/export-lora/CMakeFiles/llama-export-lora.dir/export-lora.cpp.o -o bin/llama-export-lora  common/libcommon.a  bin/libllama.so.0.0.0  bin/libggml.so.0.9.5  bin/libggml-cpu.so.0.9.5  bin/libggml-hip.so.0.9.5  bin/libggml-base.so.0.9.5  vendor/cpp-httplib/libcpp-httplib.a  /usr/lib64/libssl.so  /usr/lib64/libcrypto.so && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]
[401/401] : && /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 -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 -Xlinker --dependency-file=tools/fit-params/CMakeFiles/llama-fit-params.dir/link.d tools/fit-params/CMakeFiles/llama-fit-params.dir/fit-params.cpp.o -o bin/llama-fit-params  common/libcommon.a  bin/libllama.so.0.0.0  bin/libggml.so.0.9.5  bin/libggml-cpu.so.0.9.5  bin/libggml-hip.so.0.9.5  bin/libggml-base.so.0.9.5  vendor/cpp-httplib/libcpp-httplib.a  /usr/lib64/libssl.so  /usr/lib64/libcrypto.so && :
clang++: warning: argument unused during compilation: '-specs=/usr/lib/rpm/redhat/redhat-package-notes' [-Wunused-command-line-argument]

+ RPM_EC=0
++ jobs -p
+ exit 0
Executing(%install): /bin/sh -e /var/tmp/rpm-tmp.j8J0MW
+ umask 022
+ cd /builddir/build/BUILD/llama-cpp-b8064-build
+ '[' /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT '!=' / ']'
+ rm -rf /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT
++ dirname /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT
+ mkdir -p /builddir/build/BUILD/llama-cpp-b8064-build
+ mkdir /builddir/build/BUILD/llama-cpp-b8064-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 -mtls-dialect=gnu2 -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 -mtls-dialect=gnu2 -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 -mtls-dialect=gnu2 -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-b8064
+ DESTDIR=/builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT
+ /usr/bin/cmake --install redhat-linux-build
-- Install configuration: "Release"
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/lib64/libggml-cpu.so.0.9.5
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/lib64/libggml-cpu.so.0
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/lib64/libggml-cpu.so
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/lib64/libggml-hip.so.0.9.5
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/lib64/libggml-hip.so.0
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/lib64/libggml-hip.so
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/lib64/libggml.so.0.9.5
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/lib64/libggml.so.0
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/lib64/libggml.so
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/include/ggml.h
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/include/ggml-cpu.h
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/include/ggml-alloc.h
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/include/ggml-backend.h
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/include/ggml-blas.h
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/include/ggml-cann.h
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/include/ggml-cpp.h
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/include/ggml-cuda.h
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/include/ggml-opt.h
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/include/ggml-metal.h
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/include/ggml-rpc.h
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/include/ggml-virtgpu.h
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/include/ggml-sycl.h
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/include/ggml-vulkan.h
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/include/ggml-webgpu.h
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/include/ggml-zendnn.h
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/include/gguf.h
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/lib64/libggml-base.so.0.9.5
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/lib64/libggml-base.so.0
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/lib64/libggml-base.so
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/lib64/cmake/ggml/ggml-config.cmake
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/lib64/cmake/ggml/ggml-version.cmake
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/bin/llama-batched-bench
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/bin/llama-gguf-split
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/bin/llama-imatrix
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/bin/llama-bench
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/bin/llama-completion
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/bin/llama-perplexity
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/bin/llama-quantize
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/bin/llama-cli
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/bin/llama-server
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/bin/llama-tokenize
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/bin/llama-tts
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/lib64/libmtmd.so.0.0.0
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/lib64/libmtmd.so.0
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/lib64/libmtmd.so
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/include/mtmd.h
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/include/mtmd-helper.h
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/bin/llama-mtmd-cli
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/bin/llama-cvector-generator
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/bin/llama-export-lora
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/bin/llama-fit-params
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/lib64/libllama.so.0.0.0
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/lib64/libllama.so.0
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/lib64/libllama.so
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/include/llama.h
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/include/llama-cpp.h
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/lib64/cmake/llama/llama-config.cmake
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/lib64/cmake/llama/llama-version.cmake
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/bin/convert_hf_to_gguf.py
-- Installing: /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/lib64/pkgconfig/llama.pc
+ rm -rf '/builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/lib64/libggml_shared.*'
+ rm /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/bin/convert_hf_to_gguf.py
+ /usr/bin/find-debuginfo -j4 --strict-build-id -m -i --build-id-seed b8064-1.fc45 --unique-debug-suffix -b8064-1.fc45.x86_64 --unique-debug-src-base llama-cpp-b8064-1.fc45.x86_64 --run-dwz --dwz-low-mem-die-limit 10000000 --dwz-max-die-limit 110000000 --remove-section .gnu.build.attributes -S debugsourcefiles.list /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064
find-debuginfo: starting
Extracting debug info from 21 files
DWARF-compressing 21 files
dwz: ./usr/bin/llama-batched-bench-b8064-1.fc45.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: ./usr/bin/llama-bench-b8064-1.fc45.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: ./usr/bin/llama-cli-b8064-1.fc45.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: ./usr/bin/llama-completion-b8064-1.fc45.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: ./usr/bin/llama-cvector-generator-b8064-1.fc45.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: ./usr/bin/llama-export-lora-b8064-1.fc45.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: ./usr/bin/llama-fit-params-b8064-1.fc45.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: ./usr/bin/llama-gguf-split-b8064-1.fc45.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: ./usr/bin/llama-imatrix-b8064-1.fc45.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: ./usr/bin/llama-mtmd-cli-b8064-1.fc45.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: ./usr/bin/llama-perplexity-b8064-1.fc45.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: ./usr/bin/llama-quantize-b8064-1.fc45.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: ./usr/bin/llama-server-b8064-1.fc45.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: ./usr/bin/llama-tokenize-b8064-1.fc45.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: ./usr/bin/llama-tts-b8064-1.fc45.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: ./usr/lib64/libggml-base.so.0.9.5-b8064-1.fc45.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: ./usr/lib64/libggml-cpu.so.0.9.5-b8064-1.fc45.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: ./usr/lib64/libggml-hip.so.0.9.5-b8064-1.fc45.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: ./usr/lib64/libggml.so.0.9.5-b8064-1.fc45.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: ./usr/lib64/libllama.so.0.0.0-b8064-1.fc45.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: ./usr/lib64/libmtmd.so.0.0.0-b8064-1.fc45.x86_64.debug: Unknown debugging section .debug_str_offsets
dwz: Too few files for multifile optimization
sepdebugcrcfix: Updated 0 CRC32s, 21 CRC32s did match.
Creating .debug symlinks for symlinks to ELF files
Copying sources found by 'debugedit -l' to /usr/src/debug/llama-cpp-b8064-1.fc45.x86_64
find-debuginfo: done
+ /usr/lib/rpm/check-buildroot
+ /usr/lib/rpm/redhat/brp-ldconfig
+ COMPRESS='gzip -9 -n'
+ COMPRESS_EXT=.gz
+ /usr/lib/rpm/brp-compress
+ /usr/lib/rpm/redhat/brp-strip-lto /usr/bin/strip
+ /usr/lib/rpm/check-rpaths
+ /usr/lib/rpm/redhat/brp-mangle-shebangs
+ /usr/lib/rpm/brp-remove-la-files
+ /usr/lib/rpm/redhat/brp-python-rpm-in-distinfo
+ env /usr/lib/rpm/redhat/brp-python-bytecompile '' 1 0 -j4
+ /usr/lib/rpm/redhat/brp-python-hardlink
+ /usr/bin/add-det --brp -j4 /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT
Scanned 60 directories and 608 files,
    processed 0 inodes,
    0 modified (0 replaced + 0 rewritten),
    0 unsupported format, 0 errors
+ /usr/bin/linkdupes --brp /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr
Scanned 59 directories and 608 files,
    considered 596 files, read 134 files, linked 12 files, 0 errors
    sum of sizes of linked files: 261207 bytes
Reading /builddir/build/BUILD/llama-cpp-b8064-build/SPECPARTS/rpm-debuginfo.specpart
Processing files: llama-cpp-b8064-1.fc45.x86_64
Executing(%license): /bin/sh -e /var/tmp/rpm-tmp.z3u4e3
+ umask 022
+ cd /builddir/build/BUILD/llama-cpp-b8064-build
+ cd llama.cpp-b8064
+ LICENSEDIR=/builddir/build/BUILD/llama-cpp-b8064-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-b8064-build/BUILDROOT/usr/share/licenses/llama-cpp
+ cp -pr /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/LICENSE /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/share/licenses/llama-cpp
+ RPM_EC=0
++ jobs -p
+ exit 0
Provides: libggml-base.so.0()(64bit) libggml-cpu.so.0()(64bit) libggml-hip.so.0()(64bit) libggml.so.0()(64bit) libllama.so.0()(64bit) libmtmd.so.0()(64bit) llama-cpp = b8064-1.fc45 llama-cpp(x86-64) = b8064-1.fc45
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.7()(64bit) libamdhip64.so.7(hip_4.2)(64bit) libamdhip64.so.7(hip_6.0)(64bit) libc.so.6()(64bit) libc.so.6(GLIBC_2.10)(64bit) libc.so.6(GLIBC_2.14)(64bit) libc.so.6(GLIBC_2.15)(64bit) libc.so.6(GLIBC_2.16)(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)(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_2.42)(64bit) libc.so.6(GLIBC_2.7)(64bit) libc.so.6(GLIBC_ABI_DT_RELR)(64bit) libcrypto.so.3()(64bit) libcrypto.so.3(OPENSSL_3.0.0)(64bit) libgcc_s.so.1()(64bit) libgcc_s.so.1(GCC_12.0.0)(64bit) libgcc_s.so.1(GCC_3.0)(64bit) libggml-base.so.0()(64bit) libggml-cpu.so.0()(64bit) libggml-hip.so.0()(64bit) libggml.so.0()(64bit) libhipblas.so.3()(64bit) libllama.so.0()(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) libm.so.6(GLIBC_2.38)(64bit) libm.so.6(GLIBC_2.43)(64bit) libmtmd.so.0()(64bit) libssl.so.3()(64bit) libssl.so.3(OPENSSL_3.0.0)(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(CXXABI_1.3.7)(64bit) libstdc++.so.6(CXXABI_1.3.9)(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_3.4.32)(64bit) libstdc++.so.6(GLIBCXX_3.4.9)(64bit)
Recommends: numactl
Processing files: llama-cpp-devel-b8064-1.fc45.x86_64
Executing(%doc): /bin/sh -e /var/tmp/rpm-tmp.CHTwoL
+ umask 022
+ cd /builddir/build/BUILD/llama-cpp-b8064-build
+ cd llama.cpp-b8064
+ DOCDIR=/builddir/build/BUILD/llama-cpp-b8064-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-b8064-build/BUILDROOT/usr/share/doc/llama-cpp-devel
+ cp -pr /builddir/build/BUILD/llama-cpp-b8064-build/llama.cpp-b8064/README.md /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT/usr/share/doc/llama-cpp-devel
+ RPM_EC=0
++ jobs -p
+ exit 0
Provides: cmake(ggml) cmake(llama) llama-cpp-devel = b8064-1.fc45 llama-cpp-devel(x86-64) = b8064-1.fc45 pkgconfig(llama) = 0.0.0
Requires(rpmlib): rpmlib(CompressedFileNames) <= 3.0.4-1 rpmlib(FileDigests) <= 4.6.0-1 rpmlib(PayloadFilesHavePrefix) <= 4.0-1
Requires: /usr/bin/pkg-config cmake-filesystem(x86-64) libggml-base.so.0()(64bit) libggml-cpu.so.0()(64bit) libggml-hip.so.0()(64bit) libggml.so.0()(64bit) libllama.so.0()(64bit) libmtmd.so.0()(64bit)
Processing files: llama-cpp-debugsource-b8064-1.fc45.x86_64
Provides: llama-cpp-debugsource = b8064-1.fc45 llama-cpp-debugsource(x86-64) = b8064-1.fc45
Requires(rpmlib): rpmlib(CompressedFileNames) <= 3.0.4-1 rpmlib(FileDigests) <= 4.6.0-1 rpmlib(PayloadFilesHavePrefix) <= 4.0-1
Processing files: llama-cpp-debuginfo-b8064-1.fc45.x86_64
Provides: debuginfo(build-id) = 10adec71043b3cfefc7964a6f9d82e5c370b1d91 debuginfo(build-id) = 16ea5ec6a5a0e04805a656fb1eb3e52a61363e19 debuginfo(build-id) = 1c8299be31b7f1c844a4b0fe2464ff34ac32f2f2 debuginfo(build-id) = 224d5b00cbe427bbb8e7979be239d510bbf0d57a debuginfo(build-id) = 4230baa18dee9104e2365abba0d8194b96cb0640 debuginfo(build-id) = 5cd567199e786c4d455d6bfb3e3e6f85c79b409e debuginfo(build-id) = 739364a66b3c337768184db92da9bc37c3837b66 debuginfo(build-id) = 79ee591aab3d8b9c96c374f904ef4d217624396e debuginfo(build-id) = 916bc474df88d739e3bfdb88c1dd85daa47aad5e debuginfo(build-id) = a5de2a3c64c2aa48429e2e28374df998492c4fbc debuginfo(build-id) = a6d85e5995fd30b9ef16f8d2385974cb9c907c2c debuginfo(build-id) = a7b3b598882a10b0d98c32157709e9779ff755be debuginfo(build-id) = aab8d1adb75d0f5ea686d4f0a0837e674bfc4fa6 debuginfo(build-id) = b8ce1db3057ceccc1d3125a689bbf147d9a0fe62 debuginfo(build-id) = bd6d778538bcbb5794bc013c3fba1df40e47b371 debuginfo(build-id) = c4e31d9295addd77961d053c3c3366663b679846 debuginfo(build-id) = dc471c923430fd04bde671555cfa80f2c1e7aa2b debuginfo(build-id) = e6067849c967c50a8adec9f33d912790c36d25ad debuginfo(build-id) = e69e2ee59aa6cb13a2fede97f94e20f26e305873 debuginfo(build-id) = f75dd6a32e2c99fbdf469e026e9dea737d7a2798 debuginfo(build-id) = fe76b31ce25383b2e03880ff1153934b7ad06e3d libggml-base.so.0.9.5-b8064-1.fc45.x86_64.debug()(64bit) libggml-cpu.so.0.9.5-b8064-1.fc45.x86_64.debug()(64bit) libggml-hip.so.0.9.5-b8064-1.fc45.x86_64.debug()(64bit) libggml.so.0.9.5-b8064-1.fc45.x86_64.debug()(64bit) libllama.so.0.0.0-b8064-1.fc45.x86_64.debug()(64bit) libmtmd.so.0.0.0-b8064-1.fc45.x86_64.debug()(64bit) llama-cpp-debuginfo = b8064-1.fc45 llama-cpp-debuginfo(x86-64) = b8064-1.fc45
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) = b8064-1.fc45
Checking for unpackaged file(s): /usr/lib/rpm/check-files /builddir/build/BUILD/llama-cpp-b8064-build/BUILDROOT
Wrote: /builddir/build/RPMS/llama-cpp-devel-b8064-1.fc45.x86_64.rpm
Wrote: /builddir/build/RPMS/llama-cpp-debugsource-b8064-1.fc45.x86_64.rpm
Wrote: /builddir/build/RPMS/llama-cpp-b8064-1.fc45.x86_64.rpm
Wrote: /builddir/build/RPMS/llama-cpp-debuginfo-b8064-1.fc45.x86_64.rpm
Executing(rmbuild): /bin/sh -e /var/tmp/rpm-tmp.SEIWMx
+ umask 022
+ cd /builddir/build/BUILD/llama-cpp-b8064-build
+ test -d /builddir/build/BUILD/llama-cpp-b8064-build
+ /usr/bin/chmod -Rf a+rX,u+w,g-w,o-w /builddir/build/BUILD/llama-cpp-b8064-build
+ rm -rf /builddir/build/BUILD/llama-cpp-b8064-build
+ RPM_EC=0
++ jobs -p
+ exit 0
Finish: rpmbuild llama-cpp-b8064-1.fc45.src.rpm
Finish: build phase for llama-cpp-b8064-1.fc45.src.rpm
INFO: chroot_scan: 1 files copied to /var/lib/copr-rpmbuild/results/chroot_scan
INFO: /var/lib/mock/fedora-rawhide-x86_64-1773846985.598827/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-b8064-1.fc45.src.rpm) Config(child) 142 minutes 6 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-debuginfo",
            "epoch": null,
            "version": "b8064",
            "release": "1.fc45",
            "arch": "x86_64"
        },
        {
            "name": "llama-cpp-devel",
            "epoch": null,
            "version": "b8064",
            "release": "1.fc45",
            "arch": "x86_64"
        },
        {
            "name": "llama-cpp-debugsource",
            "epoch": null,
            "version": "b8064",
            "release": "1.fc45",
            "arch": "x86_64"
        },
        {
            "name": "llama-cpp",
            "epoch": null,
            "version": "b8064",
            "release": "1.fc45",
            "arch": "src"
        },
        {
            "name": "llama-cpp",
            "epoch": null,
            "version": "b8064",
            "release": "1.fc45",
            "arch": "x86_64"
        }
    ]
}
RPMResults finished